id
stringlengths 9
16
| submitter
stringlengths 1
64
⌀ | authors
stringlengths 3
62.8k
| title
stringlengths 1
382
| comments
stringlengths 1
1.15k
⌀ | journal-ref
stringlengths 1
557
⌀ | doi
stringlengths 7
153
⌀ | report-no
stringlengths 1
509
⌀ | categories
stringlengths 5
125
| license
stringclasses 9
values | abstract
stringlengths 6
6.09k
| versions
listlengths 1
187
| update_date
timestamp[s]date 2007-05-23 00:00:00
2025-04-25 00:00:00
| authors_parsed
sequencelengths 1
2.83k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2504.04345 | Tianxiao Huang | Tianxiao Huang, Ze Li, Jiani Liu | An abstract uncertainty principle with applications | typo fixed | null | null | null | math.AP | http://creativecommons.org/publicdomain/zero/1.0/ | Under Wigdersons' framework and by sorting out the technical points in the
recent works of Tang (J. Fourier Anal. Appl. 31 (2025)) and Dias-Luef-Prata (J.
Math. Pures Appl. (9) 198 (2025)), we prove an abstract uncertainty principle
for functions in the $L^p$ setting. An immediate consequence is a new
uncertainty principle for the Fourier transform, unifying and extending many
existing results. More applications are shown for PDEs, including the moment
growth estimates for some linear and nonlinear dispersive equations, and a type
of weighted lower bound estimate for the spacetime moment of the
Schr\"{o}dinger equation and heat equation inspired by the control theory.
| [
{
"version": "v1",
"created": "Sun, 6 Apr 2025 03:44:11 GMT"
},
{
"version": "v2",
"created": "Tue, 8 Apr 2025 14:30:01 GMT"
},
{
"version": "v3",
"created": "Thu, 10 Apr 2025 09:56:56 GMT"
},
{
"version": "v4",
"created": "Sun, 13 Apr 2025 12:34:41 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Huang",
"Tianxiao",
""
],
[
"Li",
"Ze",
""
],
[
"Liu",
"Jiani",
""
]
] |
2504.04400 | Bowen Zheng | Bowen Zheng, Enze Liu, Zhongfu Chen, Zhongrui Ma, Yue Wang, Wayne Xin
Zhao, Ji-Rong Wen | Pre-training Generative Recommender with Multi-Identifier Item
Tokenization | null | null | null | null | cs.IR cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Generative recommendation autoregressively generates item identifiers to
recommend potential items. Existing methods typically adopt a one-to-one
mapping strategy, where each item is represented by a single identifier.
However, this scheme poses issues, such as suboptimal semantic modeling for
low-frequency items and limited diversity in token sequence data. To overcome
these limitations, we propose MTGRec, which leverages Multi-identifier item
Tokenization to augment token sequence data for Generative Recommender
pre-training. Our approach involves two key innovations: multi-identifier item
tokenization and curriculum recommender pre-training. For multi-identifier item
tokenization, we leverage the RQ-VAE as the tokenizer backbone and treat model
checkpoints from adjacent training epochs as semantically relevant tokenizers.
This allows each item to be associated with multiple identifiers, enabling a
single user interaction sequence to be converted into several token sequences
as different data groups. For curriculum recommender pre-training, we introduce
a curriculum learning scheme guided by data influence estimation, dynamically
adjusting the sampling probability of each data group during recommender
pre-training. After pre-training, we fine-tune the model using a single
tokenizer to ensure accurate item identification for recommendation. Extensive
experiments on three public benchmark datasets demonstrate that MTGRec
significantly outperforms both traditional and generative recommendation
baselines in terms of effectiveness and scalability.
| [
{
"version": "v1",
"created": "Sun, 6 Apr 2025 08:03:03 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 02:51:58 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zheng",
"Bowen",
""
],
[
"Liu",
"Enze",
""
],
[
"Chen",
"Zhongfu",
""
],
[
"Ma",
"Zhongrui",
""
],
[
"Wang",
"Yue",
""
],
[
"Zhao",
"Wayne Xin",
""
],
[
"Wen",
"Ji-Rong",
""
]
] |
2504.04401 | Zhou Youyang | Zhou Youyang, Shi Weiren, Xie Yun, Zhao Bianli, Luo Xinyu, Yao
Mingjie, Zhang Rui, Tan Xin, Li Kui, Yang Hao, Liu Qi, Nan Yinggang, Bao Jie,
Zhang Yuping, Shu Feng, Li Shaopan and Zhang Xiaoshi | Super-Resolution Coherent Diffractive Imaging via Titled-Incidence
Multi-Rotation-Angle Fusion Ptychography | 18 pages, 6 figures | null | null | null | physics.optics | http://creativecommons.org/licenses/by/4.0/ | Coherent diffractive imaging (CDI) enables lensless imaging with experimental
simplicity and a flexible field of view, yet its resolution is fundamentally
constrained by the Abbe diffraction limit. To overcome this limitation, we
introduce a novel Tilted-Incidence Multi-Rotation-Angle Fusion Ptychography
technique. This approach leverages a tilted-incidence geometry to extend the
collection angle beyond the Abbe limit, achieving up to a -fold resolution
enhancement. By acquiring diffraction patterns at multiple sample rotation
angles, we capture complementary spatial frequency information. A
tilted-incidence multi-rotation-angle fusion ptychographic iterative engine
(tmf-PIE) algorithm is then employed to integrate these datasets, enabling
super-resolution image reconstruction. Additionally, this method mitigates the
anisotropic resolution artifacts inherent to tilted CDI geometries. Our
technique represents a novel advancement in super-resolution imaging, providing
a novel alternative alongside established methods such as STED, SIM, and SMLM.
| [
{
"version": "v1",
"created": "Sun, 6 Apr 2025 08:03:43 GMT"
},
{
"version": "v2",
"created": "Tue, 8 Apr 2025 08:24:46 GMT"
},
{
"version": "v3",
"created": "Sun, 13 Apr 2025 06:44:08 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Youyang",
"Zhou",
""
],
[
"Weiren",
"Shi",
""
],
[
"Yun",
"Xie",
""
],
[
"Bianli",
"Zhao",
""
],
[
"Xinyu",
"Luo",
""
],
[
"Mingjie",
"Yao",
""
],
[
"Rui",
"Zhang",
""
],
[
"Xin",
"Tan",
""
],
[
"Kui",
"Li",
""
],
[
"Hao",
"Yang",
""
],
[
"Qi",
"Liu",
""
],
[
"Yinggang",
"Nan",
""
],
[
"Jie",
"Bao",
""
],
[
"Yuping",
"Zhang",
""
],
[
"Feng",
"Shu",
""
],
[
"Shaopan",
"Li",
""
],
[
"Xiaoshi",
"Zhang",
""
]
] |
2504.04405 | Bowen Zheng | Bowen Zheng, Hongyu Lu, Yu Chen, Wayne Xin Zhao, Ji-Rong Wen | Universal Item Tokenization for Transferable Generative Recommendation | null | null | null | null | cs.IR cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, generative recommendation has emerged as a promising paradigm,
attracting significant research attention. The basic framework involves an item
tokenizer, which represents each item as a sequence of codes serving as its
identifier, and a generative recommender that predicts the next item by
autoregressively generating the target item identifier. However, in existing
methods, both the tokenizer and the recommender are typically domain-specific,
limiting their ability for effective transfer or adaptation to new domains. To
this end, we propose UTGRec, a Universal item Tokenization approach for
transferable Generative Recommendation. Specifically, we design a universal
item tokenizer for encoding rich item semantics by adapting a multimodal large
language model (MLLM). By devising tree-structured codebooks, we discretize
content representations into corresponding codes for item tokenization. To
effectively learn the universal item tokenizer on multiple domains, we
introduce two key techniques in our approach. For raw content reconstruction,
we employ dual lightweight decoders to reconstruct item text and images from
discrete representations to capture general knowledge embedded in the content.
For collaborative knowledge integration, we assume that co-occurring items are
similar and integrate collaborative signals through co-occurrence alignment and
reconstruction. Finally, we present a joint learning framework to pre-train and
adapt the transferable generative recommender across multiple domains.
Extensive experiments on four public datasets demonstrate the superiority of
UTGRec compared to both traditional and generative recommendation baselines.
| [
{
"version": "v1",
"created": "Sun, 6 Apr 2025 08:07:49 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 02:50:20 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zheng",
"Bowen",
""
],
[
"Lu",
"Hongyu",
""
],
[
"Chen",
"Yu",
""
],
[
"Zhao",
"Wayne Xin",
""
],
[
"Wen",
"Ji-Rong",
""
]
] |
2504.04430 | Matej \v{S}progar | Matej \v{S}progar | AGITB: A Signal-Level Benchmark for Evaluating Artificial General
Intelligence | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Despite remarkable progress in machine learning, current AI systems continue
to fall short of true human-like intelligence. While Large Language Models
(LLMs) excel in pattern recognition and response generation, they lack genuine
understanding - an essential hallmark of Artificial General Intelligence (AGI).
Existing AGI evaluation methods fail to offer a practical, gradual, and
informative metric. This paper introduces the Artificial General Intelligence
Test Bed (AGITB), comprising twelve rigorous tests that form a
signal-processing-level foundation for the potential emergence of cognitive
capabilities. AGITB evaluates intelligence through a model's ability to predict
binary signals across time without relying on symbolic representations or
pretraining. Unlike high-level tests grounded in language or perception, AGITB
focuses on core computational invariants reflective of biological intelligence,
such as determinism, sensitivity, and generalisation. The test bed assumes no
prior bias, operates independently of semantic meaning, and ensures
unsolvability through brute force or memorization. While humans pass AGITB by
design, no current AI system has met its criteria, making AGITB a compelling
benchmark for guiding and recognizing progress toward AGI.
| [
{
"version": "v1",
"created": "Sun, 6 Apr 2025 10:01:15 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 10:03:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Šprogar",
"Matej",
""
]
] |
2504.04450 | Lu Bai | Lu Bai, Mengtong Li, Siyuan Lian, Kai Chen, Jing Lu | WaveNet-Volterra Neural Networks for Active Noise Control: A Fully
Causal Approach | null | null | null | null | eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Active Noise Control (ANC) systems are challenged by nonlinear distortions,
which degrade the performance of traditional adaptive filters. While deep
learning-based ANC algorithms have emerged to address nonlinearity, existing
approaches often overlook critical limitations: (1) end-to-end Deep Neural
Network (DNN) models frequently violate causality constraints inherent to
real-time ANC applications; (2) many studies compare DNN-based methods against
simplified or low-order adaptive filters rather than fully optimized high-order
counterparts. In this letter, we propose a causality-preserving time-domain ANC
framework that synergizes WaveNet with Volterra Neural Networks (VNNs),
explicitly addressing system nonlinearity while ensuring strict causal
operation. Unlike prior DNN-based approaches, our method is benchmarked against
both state-of-the-art deep learning architectures and rigorously optimized
high-order adaptive filters, including Wiener solutions. Simulations
demonstrate that the proposed framework achieves superior performance over
existing DNN methods and traditional algorithms, revealing that prior claims of
DNN superiority stem from incomplete comparisons with suboptimal traditional
baselines. Source code is available at
https://github.com/Lu-Baihh/WaveNet-VNNs-for-ANC.git.
| [
{
"version": "v1",
"created": "Sun, 6 Apr 2025 11:42:01 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 02:51:20 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Bai",
"Lu",
""
],
[
"Li",
"Mengtong",
""
],
[
"Lian",
"Siyuan",
""
],
[
"Chen",
"Kai",
""
],
[
"Lu",
"Jing",
""
]
] |
2504.04569 | Chitranshu Harbola | Chitranshu Harbola and Anupam Purwar | KnowsLM: A framework for evaluation of small language models for
knowledge augmentation and humanised conversations | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | In the evolving landscape of conversational AI, generating concise,
context-aware, and human-like dialogue using small and medium-sized language
models (LLMs) remains a complex challenge. This study investigates the
influence of LoRA rank, dataset scale, and prompt prefix design on both
knowledge retention and stylistic alignment. While fine-tuning improves fluency
and enables stylistic customization, its ability to integrate unseen knowledge
is constrained -- particularly with smaller datasets. Conversely, RAG-augmented
models, equipped to incorporate external documents at inference, demonstrated
superior factual accuracy on out-of-distribution prompts, though they lacked
the stylistic consistency achieved by fine-tuning. Evaluations by LLM-based
judges across knowledge accuracy, conversational quality, and conciseness
suggest that fine-tuning is best suited for tone adaptation, whereas RAG excels
at real-time knowledge augmentation.
| [
{
"version": "v1",
"created": "Sun, 6 Apr 2025 17:58:08 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 14:53:45 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Harbola",
"Chitranshu",
""
],
[
"Purwar",
"Anupam",
""
]
] |
2504.04603 | Pau Marquez Julbe | Pau Marquez Julbe, Julian Nubert, Henrik Hose, Sebastian Trimpe,
Katherine J. Kuchenbecker | Diffusion-Based Approximate MPC: Fast and Consistent Imitation of
Multi-Modal Action Distributions | null | null | null | null | cs.RO | http://creativecommons.org/licenses/by/4.0/ | Approximating model predictive control (MPC) using imitation learning (IL)
allows for fast control without solving expensive optimization problems online.
However, methods that use neural networks in a simple L2-regression setup fail
to approximate multi-modal (set-valued) solution distributions caused by local
optima found by the numerical solver or non-convex constraints, such as
obstacles, significantly limiting the applicability of approximate MPC in
practice. We solve this issue by using diffusion models to accurately represent
the complete solution distribution (i.e., all modes) at high control rates
(more than 1000 Hz). This work shows that diffusion based AMPC significantly
outperforms L2-regression-based approximate MPC for multi-modal action
distributions. In contrast to most earlier work on IL, we also focus on running
the diffusion-based controller at a higher rate and in joint space instead of
end-effector space. Additionally, we propose the use of gradient guidance
during the denoising process to consistently pick the same mode in closed loop
to prevent switching between solutions. We propose using the cost and
constraint satisfaction of the original MPC problem during parallel sampling of
solutions from the diffusion model to pick a better mode online. We evaluate
our method on the fast and accurate control of a 7-DoF robot manipulator both
in simulation and on hardware deployed at 250 Hz, achieving a speedup of more
than 70 times compared to solving the MPC problem online and also outperforming
the numerical optimization (used for training) in success ratio. Project
website: https://paumarquez.github.io/diffusion-ampc.
| [
{
"version": "v1",
"created": "Sun, 6 Apr 2025 20:22:01 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 11:05:39 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Julbe",
"Pau Marquez",
""
],
[
"Nubert",
"Julian",
""
],
[
"Hose",
"Henrik",
""
],
[
"Trimpe",
"Sebastian",
""
],
[
"Kuchenbecker",
"Katherine J.",
""
]
] |
2504.04637 | Stanislav Semenov | Stanislav Semenov | On the Nature of Fractal Numbers and the Classical Continuum Hypothesis
(CH) | 35 pages, submitted to arXiv | null | null | null | math.LO cs.LO | http://creativecommons.org/licenses/by/4.0/ | We propose a reinterpretation of the continuum grounded in the stratified
structure of definability rather than classical cardinality. In this framework,
a real number is not an abstract point on the number line, but an object
expressible at some level Fn of a formal hierarchy. We introduce the notion of
"fractal numbers" -- entities defined not within a fixed set-theoretic
universe, but through layered expressibility across constructive systems. This
reconceptualizes irrationality as a relative property, depending on
definability depth, and replaces the binary dichotomy between countable and
uncountable sets with a gradated spectrum of definability classes. We show that
the classical Continuum Hypothesis loses its force in this context: between
aleph_0 and c lies not a single cardinal jump, but a stratified sequence of
definitional stages, each forming a countable yet irreducible approximation to
the continuum. We argue that the real line should not be seen as a completed
totality but as an evolving architecture of formal expressibility. We conclude
with a discussion of rational invariants, the relativity of irrationality, and
the emergence of a fractal metric for definitional density.
| [
{
"version": "v1",
"created": "Sun, 6 Apr 2025 22:40:21 GMT"
},
{
"version": "v2",
"created": "Tue, 8 Apr 2025 09:52:56 GMT"
},
{
"version": "v3",
"created": "Wed, 9 Apr 2025 15:39:02 GMT"
},
{
"version": "v4",
"created": "Thu, 10 Apr 2025 12:46:52 GMT"
},
{
"version": "v5",
"created": "Sun, 13 Apr 2025 17:28:39 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Semenov",
"Stanislav",
""
]
] |
2504.04639 | Alberto Larrauri | Alberto Larrauri | Ineffectiveness for Search and Undecidability of PCSP Meta-Problems | null | null | null | null | cs.CC cs.CL cs.DS cs.LO | http://creativecommons.org/licenses/by/4.0/ | It is an open question whether the search and decision versions of promise
CSPs are equivalent. Most known algorithms for PCSPs solve only their
\emph{decision} variant, and it is unknown whether they can be adapted to solve
\emph{search} as well. The main approaches, called BLP, AIP and BLP+AIP, handle
a PCSP by finding a solution to a relaxation of some integer program. We prove
that rounding those solutions to a proper search certificate can be as hard as
any problem in the class TFNP. In other words, these algorithms are ineffective
for search. Building on the algebraic approach to PCSPs, we find sufficient
conditions that imply ineffectiveness for search. Our tools are tailored to
algorithms that are characterized by minions in a suitable way, and can also be
used to prove undecidability results for meta-problems. This way, we show that
the families of templates solvable via BLP, AIP, and BLP+AIP are undecidable.
Using the same techniques we also analyze several algebraic conditions that
are known to guarantee the tractability of finite-template CSPs. We prove that
several meta-problems related to cyclic polymorphims and WNUs are undecidable
for PCSPs. In particular, there is no algorithm deciding whether a finite PCSP
template (1) admits cyclic a polymorphism, (2) admits a WNU.
| [
{
"version": "v1",
"created": "Sun, 6 Apr 2025 23:08:05 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 23:31:22 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Larrauri",
"Alberto",
""
]
] |
2504.04672 | HuiChang Li | Hui Chang Li, Yun Shen and Xiao Hua Deng | Topological Anderson Phase Transitions in Y-shaped Plasmonic Valley
Metal-slabs | 6 pages, 4 figures | null | null | null | physics.optics | http://creativecommons.org/licenses/by/4.0/ | Throughout history, all developmental trajectories of civilization -
encompassing progress, creation, and innovation - have fundamentally pursued
the paradigm shift 'from disorder to order'. In photonics, investigations into
disordered systems have primarily focused on foundational principles governing
signal diffusion and localization. This paper addresses terahertz device
development by examining the dual role of disorder in photonic systems: while
potentially compromising optical transmission stability, it simultaneously
inspires innovative topological protection mechanisms. Building upon the
symmetry-breaking induced valley-Hall topological Anderson phase transition in
Y-shaped metallic structures, we achieve valley Chern number modulation through
random rotation of constituent units, demonstrating progressive emergence of
in-gap topological states with increasing disorder parameters and observing
topological negative refraction phenomena. Furthermore, an effective Dirac
two-band model is established to quantitatively characterize the evolution of
bulk transport states under disorder variation. By strategically regulating
disordered configurations to induce valley-Hall topological Anderson phase
transitions, this research provides new pathways for overcoming critical
technical challenges in terahertz devices, particularly transmission loss
limitations.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 01:46:59 GMT"
},
{
"version": "v2",
"created": "Thu, 10 Apr 2025 06:56:07 GMT"
},
{
"version": "v3",
"created": "Mon, 14 Apr 2025 12:36:28 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Li",
"Hui Chang",
""
],
[
"Shen",
"Yun",
""
],
[
"Deng",
"Xiao Hua",
""
]
] |
2504.04695 | Razieh Emami Meibody | Razieh Emami, Matthew Liska, Koushik Chatterjee, Geoffrey C. Bower,
Wystan Benbow, Douglas Finkbeiner, Maciek Wielgus, Lars Hernquist, Randall
Smith, Grant Tremblay, Angelo Ricarte, James F. Steiner, Avery E. Broderick,
Saurabh, Jordy Davelaar, Josh Grindlay, Mark Vogelsberger, and Chi-Kwan Chan | Novel Polarimetric Analysis of Near Horizon Flaring Episodes in M87* in
Millimeter Wavelength | 24 pages, 23 figures | null | null | null | astro-ph.HE astro-ph.GA hep-ph hep-th | http://creativecommons.org/licenses/by/4.0/ | Recent multi-wavelength observations of M87* \citep{2024A&A...692A.140A}
revealed a high-energy $\gamma$-ray flare without a corresponding millimeter
counterpart. We present a theoretical polarimetric study to evaluate the
presence and nature of a potential millimeter flare in M87*, using a suite of
general relativistic magnetohydrodynamical simulations with varying black hole
(BH) spins and magnetic field configurations. We find that the emergence of a
millimeter flare is strongly influenced by both spin and magnetic structure,
with limited sensitivity to the electron distribution (thermal vs.
non-thermal). We model the intensity light curve with a damped random walk
(DRW) and compare the characteristic timescale ($\tau$) with recent SMA
observations, finding that the simulated $\tau$ exceeds observed values by over
an order of magnitude. In a flaring case with BH spin a=+0.5, we identify a
distinct millimeter flare followed by an order-of-magnitude flux drop. All
Stokes parameters show variability near the flare, including a sign reversal in
the electric vector position angle. While most $\beta_m$ modes remain stable,
the $EB$-correlation phase is highly sensitive to both the flare peak and
decay. We examine polarimetric signatures in photon sub-rings, focusing on
modes ns=0 and ns=1. The ns=0 signal closely matches the full image, while ns=1
reveals distinct behaviors, highlighting the potential of space VLBI to isolate
sub-ring features. Finally, we analyze the magnetic and velocity field
evolution during the flare, finding that magnetic reconnection weakens during
the flux decay, and the clockwise velocity flow transitions into an
outflow-dominated regime. These results suggest that transient radio
variability near flares encodes key information about black hole spin and
magnetic field structure, offering a novel probe into the physics of active
galactic nuclei.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 03:00:21 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 17:37:12 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Emami",
"Razieh",
""
],
[
"Liska",
"Matthew",
""
],
[
"Chatterjee",
"Koushik",
""
],
[
"Bower",
"Geoffrey C.",
""
],
[
"Benbow",
"Wystan",
""
],
[
"Finkbeiner",
"Douglas",
""
],
[
"Wielgus",
"Maciek",
""
],
[
"Hernquist",
"Lars",
""
],
[
"Smith",
"Randall",
""
],
[
"Tremblay",
"Grant",
""
],
[
"Ricarte",
"Angelo",
""
],
[
"Steiner",
"James F.",
""
],
[
"Broderick",
"Avery E.",
""
],
[
"Saurabh",
"",
""
],
[
"Davelaar",
"Jordy",
""
],
[
"Grindlay",
"Josh",
""
],
[
"Vogelsberger",
"Mark",
""
],
[
"Chan",
"Chi-Kwan",
""
]
] |
2504.04833 | Andrea Esposito | Andrea Esposito (1), Miriana Calvano (1), Antonio Curci (1 and 2),
Francesco Greco (1), Rosa Lanzilotti (1), Antonio Piccinno (1) ((1)
University of Bari Aldo Moro, (2) University of Pisa) | Explanation-Driven Interventions for Artificial Intelligence Model
Customization: Empowering End-Users to Tailor Black-Box AI in Rhinocytology | Second version (11 pages, 8 of content) | null | null | null | cs.HC cs.AI | http://creativecommons.org/licenses/by/4.0/ | The integration of Artificial Intelligence (AI) in modern society is
transforming how individuals perform tasks. In high-risk domains, ensuring
human control over AI systems remains a key design challenge. This article
presents a novel End-User Development (EUD) approach for black-box AI models,
enabling users to edit explanations and influence future predictions through
targeted interventions. By combining explainability, user control, and model
adaptability, the proposed method advances Human-Centered AI (HCAI), promoting
a symbiotic relationship between humans and adaptive, user-tailored AI systems.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 08:44:48 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 16:21:20 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Esposito",
"Andrea",
"",
"1 and 2"
],
[
"Calvano",
"Miriana",
"",
"1 and 2"
],
[
"Curci",
"Antonio",
"",
"1 and 2"
],
[
"Greco",
"Francesco",
""
],
[
"Lanzilotti",
"Rosa",
""
],
[
"Piccinno",
"Antonio",
""
]
] |
2504.04840 | Zhaofeng Shi | Zhaofeng Shi, Heqian Qiu, Lanxiao Wang, Qingbo Wu, Fanman Meng, and
Hongliang Li | Unsupervised Ego- and Exo-centric Dense Procedural Activity Captioning
via Gaze Consensus Adaptation | null | null | null | null | cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Even from an early age, humans naturally adapt between exocentric (Exo) and
egocentric (Ego) perspectives to understand daily procedural activities.
Inspired by this cognitive ability, we propose a novel Unsupervised Ego-Exo
Dense Procedural Activity Captioning (UE$^{2}$DPAC) task, which aims to
transfer knowledge from the labeled source view to predict the time segments
and descriptions of action sequences for the target view without annotations.
Despite previous works endeavoring to address the fully-supervised single-view
or cross-view dense video captioning, they lapse in the proposed task due to
the significant inter-view gap caused by temporal misalignment and irrelevant
object interference. Hence, we propose a Gaze Consensus-guided Ego-Exo
Adaptation Network (GCEAN) that injects the gaze information into the learned
representations for the fine-grained Ego-Exo alignment. Specifically, we
propose a Score-based Adversarial Learning Module (SALM) that incorporates a
discriminative scoring network and compares the scores of distinct views to
learn unified view-invariant representations from a global level. Then, the
Gaze Consensus Construction Module (GCCM) utilizes the gaze to progressively
calibrate the learned representations to highlight the regions of interest and
extract the corresponding temporal contexts. Moreover, we adopt hierarchical
gaze-guided consistency losses to construct gaze consensus for the explicit
temporal and spatial adaptation between the source and target views. To support
our research, we propose a new EgoMe-UE$^{2}$DPAC benchmark, and extensive
experiments demonstrate the effectiveness of our method, which outperforms many
related methods by a large margin. The code will be released.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 08:51:11 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 04:16:24 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Shi",
"Zhaofeng",
""
],
[
"Qiu",
"Heqian",
""
],
[
"Wang",
"Lanxiao",
""
],
[
"Wu",
"Qingbo",
""
],
[
"Meng",
"Fanman",
""
],
[
"Li",
"Hongliang",
""
]
] |
2504.04843 | Yizhou Dang | Yizhou Dang, Yuting Liu, Enneng Yang, Minhan Huang, Guibing Guo,
Jianzhe Zhao, Xingwei Wang | Data Augmentation as Free Lunch: Exploring the Test-Time Augmentation
for Sequential Recommendation | Accepted by SIGIR 2025 Full Paper | null | null | null | cs.IR | http://creativecommons.org/licenses/by/4.0/ | Data augmentation has become a promising method of mitigating data sparsity
in sequential recommendation. Existing methods generate new yet effective data
during model training to improve performance. However, deploying them requires
retraining, architecture modification, or introducing additional learnable
parameters. The above steps are time-consuming and costly for well-trained
models, especially when the model scale becomes large. In this work, we explore
the test-time augmentation (TTA) for sequential recommendation, which augments
the inputs during the model inference and then aggregates the model's
predictions for augmented data to improve final accuracy. It avoids significant
time and cost overhead from loss calculation and backward propagation. We first
experimentally disclose the potential of existing augmentation operators for
TTA and find that the Mask and Substitute consistently achieve better
performance. Further analysis reveals that these two operators are effective
because they retain the original sequential pattern while adding appropriate
perturbations. Meanwhile, we argue that these two operators still face
time-consuming item selection or interference information from mask tokens.
Based on the analysis and limitations, we present TNoise and TMask. The former
injects uniform noise into the original representation, avoiding the
computational overhead of item selection. The latter blocks mask token from
participating in model calculations or directly removes interactions that
should have been replaced with mask tokens. Comprehensive experiments
demonstrate the effectiveness, efficiency, and generalizability of our method.
We provide an anonymous implementation at https://github.com/KingGugu/TTA4SR.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 08:56:16 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 06:01:59 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Dang",
"Yizhou",
""
],
[
"Liu",
"Yuting",
""
],
[
"Yang",
"Enneng",
""
],
[
"Huang",
"Minhan",
""
],
[
"Guo",
"Guibing",
""
],
[
"Zhao",
"Jianzhe",
""
],
[
"Wang",
"Xingwei",
""
]
] |
2504.04867 | Yan-Ann Chen | Ming-Lun Lee, Han-Chang Chou, Yan-Ann Chen | FedSAUC: A Similarity-Aware Update Control for Communication-Efficient
Federated Learning in Edge Computing | Published in the Proceedings of the International Conference on
Mobile Computing and Ubiquitous Network (ICMU), 2021 | null | 10.23919/ICMU50196.2021.9638814 | null | cs.LG cs.AI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Federated learning is a distributed machine learning framework to
collaboratively train a global model without uploading privacy-sensitive data
onto a centralized server. Usually, this framework is applied to edge devices
such as smartphones, wearable devices, and Internet of Things (IoT) devices
which closely collect information from users. However, these devices are mostly
battery-powered. The update procedure of federated learning will constantly
consume the battery power and the transmission bandwidth. In this work, we
propose an update control for federated learning, FedSAUC, by considering the
similarity of users' behaviors (models). At the server side, we exploit
clustering algorithms to group devices with similar models. Then we select some
representatives for each cluster to update information to train the model. We
also implemented a testbed prototyping on edge devices for validating the
performance. The experimental results show that this update control will not
affect the training accuracy in the long run.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 09:21:43 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Lee",
"Ming-Lun",
""
],
[
"Chou",
"Han-Chang",
""
],
[
"Chen",
"Yan-Ann",
""
]
] |
2504.04984 | Pawe{\l} Rz\k{a}\.zewski | Nadzieja Hodur, Monika Pil\'sniak, Magdalena Prorok, Pawe{\l}
Rz\k{a}\.zewski | Finding large $k$-colorable induced subgraphs in (bull, chair)-free and
(bull,E)-free graphs | null | null | null | null | cs.CC math.CO | http://creativecommons.org/licenses/by/4.0/ | We study the Max Partial $k$-Coloring problem, where we are given a
vertex-weighted graph, and we ask for a maximum-weight induced subgraph that
admits a proper $k$-coloring. For $k=1$ this problem coincides with Maximum
Weight Independent Set, and for $k=2$ the problem is equivalent (by
complementation) to Minimum Odd Cycle Transversal. Furthermore, it generalizes
$k$-Coloring. We show that Max Partial $k$-Coloring on $n$-vertex instances
with clique number $\omega$ can be solved in time
* $n^{\mathcal{O}(k\omega)}$ if the input graph excludes the bull and the
chair as an induced subgraph,
* $n^{\mathcal{O}(k\omega \log n)}$ if the input graph excludes the bull and
E as an induced subgraph.
This implies that $k$-Coloring can be solved in polynomial time in the former
class, and in quasipolynomial-time in the latter one.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 12:11:01 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hodur",
"Nadzieja",
""
],
[
"Pilśniak",
"Monika",
""
],
[
"Prorok",
"Magdalena",
""
],
[
"Rzążewski",
"Paweł",
""
]
] |
2504.05005 | Jonathan Brady | Jonathan Brady, Vijay Chandiramani, Michelle Chatwin, Emmanuel Lwele,
Aminat Yetunde Saula, Peter Stewart, Toby Stokes | Can audio recordings be used to detect leaks and coughs during
mechanical insufflation exsufflation (MI-E) treatment? | null | null | null | null | physics.med-ph | http://creativecommons.org/licenses/by/4.0/ | This report relates to a study group hosted by the EPSRC funded network,
Integrating data-driven BIOphysical models into REspiratory MEdicine (BIOREME),
and supported by SofTMech and Innovate UK, Business Connect. The BIOREME
network hosts events, including this study group, to bring together
multi-disciplinary researchers, clinicians, companies and charities to catalyse
research in the applications of mathematical modelling for respiratory
medicine. The goal of this study group was to provide an interface between
companies, clinicians, and mathematicians to develop mathematical tools to the
problems presented. The study group was held at The University of Glasgow on
the 17 - 21 June 2024 and was attended by 16 participants from 8 different
institutions. Below details the technical report of one of the challenges and
the methods developed by the team of researchers who worked on this challenge.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 12:32:01 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 09:32:03 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Brady",
"Jonathan",
""
],
[
"Chandiramani",
"Vijay",
""
],
[
"Chatwin",
"Michelle",
""
],
[
"Lwele",
"Emmanuel",
""
],
[
"Saula",
"Aminat Yetunde",
""
],
[
"Stewart",
"Peter",
""
],
[
"Stokes",
"Toby",
""
]
] |
2504.05035 | Joseph Kibugi Chege | Joseph K. Chege, Arie Yeredor, Martin Haardt | Probabilistic Position-Aided Beam Selection for mmWave MIMO Systems | null | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | Millimeter-wave (mmWave) MIMO systems rely on highly directional beamforming
to overcome severe path loss and ensure robust communication links. However,
selecting the optimal beam pair efficiently remains a challenge due to the
large search space and the overhead of conventional methods. This paper
proposes a probabilistic position-aided beam selection approach that exploits
the statistical dependence between user equipment (UE) positions and optimal
beam indices. We model the underlying joint probability mass function (PMF) of
the positions and the beam indices as a low-rank tensor and estimate its
parameters from training data using Bayesian inference. The estimated model is
then used to predict the best (or a list of the top) beam pair indices for new
UE positions. The proposed method is evaluated using data generated from a
state-of-the-art ray tracing simulator and compared with neural network-based
and fingerprinting approaches. The results show that our approach achieves a
high data rate with fewer training samples and a significantly reduced beam
search space. These advantages render it a promising solution for practical
mmWave MIMO deployments, reducing the beam search overhead while maintaining a
reliable connectivity.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 12:56:51 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 19:25:29 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Chege",
"Joseph K.",
""
],
[
"Yeredor",
"Arie",
""
],
[
"Haardt",
"Martin",
""
]
] |
2504.05045 | Zhikun Yang | Huilin Yin, Zhikun Yang, Linchuan Zhang, Daniel Watzenig | Attention-Augmented Inverse Reinforcement Learning with Graph
Convolutions for Multi-Agent Task Allocation | This version includes changes made to meet the submission
requirements of IEEE Transactions on Vehicular Technology (TVT): author
biographies and IEEE copyright footer removed; acknowledgment anonymized;
author footnotes updated; a co-author added for figure illustration and minor
edits | null | null | null | cs.LG cs.MA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work has been submitted to the IEEE for possible publication. Copyright
may be transferred without notice, after which this version may no longer be
accessible.
Multi-agent task allocation (MATA) plays a vital role in cooperative
multi-agent systems, with significant implications for applications such as
logistics, search and rescue, and robotic coordination. Although traditional
deep reinforcement learning (DRL) methods have been shown to be promising,
their effectiveness is hindered by a reliance on manually designed reward
functions and inefficiencies in dynamic environments. In this paper, an inverse
reinforcement learning (IRL)-based framework is proposed, in which multi-head
self-attention (MHSA) and graph attention mechanisms are incorporated to
enhance reward function learning and task execution efficiency. Expert
demonstrations are utilized to infer optimal reward densities, allowing
dependence on handcrafted designs to be reduced and adaptability to be
improved. Extensive experiments validate the superiority of the proposed method
over widely used multi-agent reinforcement learning (MARL) algorithms in terms
of both cumulative rewards and task execution efficiency.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 13:14:45 GMT"
},
{
"version": "v2",
"created": "Tue, 8 Apr 2025 03:33:08 GMT"
},
{
"version": "v3",
"created": "Mon, 14 Apr 2025 09:42:55 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Yin",
"Huilin",
""
],
[
"Yang",
"Zhikun",
""
],
[
"Zhang",
"Linchuan",
""
],
[
"Watzenig",
"Daniel",
""
]
] |
2504.05083 | Lekshmi Dathan | Lekshmi Dathan and Pankaj Kushwaha | X-ray variability of VHE detected FSRQs: A comparative study | null | null | null | null | astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Flat Spectrum Radio Quasars (FSRQs) are weak sources of very high energy
(VHE; E>100 GeV) emission, despite exhibiting strong MeV-GeV emissions that
dominate their radiative output. To date, only ten FSRQs have been detected at
VHEs, primarily during bright optical phases. In this study, we perform a
detailed and systematic, temporal, and spectral analysis of the nine
VHE-detected FSRQs, using the Swift X-ray Telescope (XRT) data. Our findings
show no correlation between VHE activity and the X-ray flux or spectral state
of the sources. However, investigation of spectral properties with X-ray
brightness shows anti-correlation between flux and spectral index. The X-ray,
generally with a different spectral shape lies at the farther end of the
optical-UV synchrotron spectrum which typically shows a declining power-law
spectrum, and thus, the X-ray spectrum is generally explained by Synchrotron
Self-Compton (SSC) process. However, if optical-UV synchrotron emission extends
into the X-ray band, it can soften the X-ray spectrum. While most sources in
our sample exhibit rising X-ray SEDs, indicative of non-synchrotron origins or
minimal synchrotron contributions, many display softer or flat X-ray spectra,
mainly during low X-ray flux states (e.g., 4C +21.35, 3C 279, TON 0599, PKS
1441+25, and PKS 0346-27) suggesting potential synchrotron contributions. These
synchrotron continuations influence the gamma-ray spectrum, implying extension
into the VHE range for inverse Compton (IC) scattering in the Thomson
scattering limit. If the extended component corresponds to an underlying
low-level emission, these FSRQs could represent potential candidates for
persistent VHE activity.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 13:53:57 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 08:08:48 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Dathan",
"Lekshmi",
""
],
[
"Kushwaha",
"Pankaj",
""
]
] |
2504.05089 | Johannes Dollinger | Johannes Dollinger, Damien Robert, Elena Plekhanova, Lukas Drees, Jan
Dirk Wegner | Climplicit: Climatic Implicit Embeddings for Global Ecological Tasks | Published as a workshop paper at "Tackling Climate Change with
Machine Learning", ICLR 2025 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Deep learning on climatic data holds potential for macroecological
applications. However, its adoption remains limited among scientists outside
the deep learning community due to storage, compute, and technical expertise
barriers. To address this, we introduce Climplicit, a spatio-temporal
geolocation encoder pretrained to generate implicit climatic representations
anywhere on Earth. By bypassing the need to download raw climatic rasters and
train feature extractors, our model uses x3500 less disk space and
significantly reduces computational needs for downstream tasks. We evaluate our
Climplicit embeddings on biomes classification, species distribution modeling,
and plant trait regression. We find that single-layer probing our Climplicit
embeddings consistently performs better or on par with training a model from
scratch on downstream tasks and overall better than alternative geolocation
encoding models.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 13:58:55 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 12:47:23 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Dollinger",
"Johannes",
""
],
[
"Robert",
"Damien",
""
],
[
"Plekhanova",
"Elena",
""
],
[
"Drees",
"Lukas",
""
],
[
"Wegner",
"Jan Dirk",
""
]
] |
2504.05108 | Anja Surina | Anja Surina, Amin Mansouri, Lars Quaedvlieg, Amal Seddas, Maryna
Viazovska, Emmanuel Abbe, Caglar Gulcehre | Algorithm Discovery With LLMs: Evolutionary Search Meets Reinforcement
Learning | 29 pages | null | null | null | cs.AI cs.LG cs.NE | http://creativecommons.org/licenses/by/4.0/ | Discovering efficient algorithms for solving complex problems has been an
outstanding challenge in mathematics and computer science, requiring
substantial human expertise over the years. Recent advancements in evolutionary
search with large language models (LLMs) have shown promise in accelerating the
discovery of algorithms across various domains, particularly in mathematics and
optimization. However, existing approaches treat the LLM as a static generator,
missing the opportunity to update the model with the signal obtained from
evolutionary exploration. In this work, we propose to augment LLM-based
evolutionary search by continuously refining the search operator - the LLM -
through reinforcement learning (RL) fine-tuning. Our method leverages
evolutionary search as an exploration strategy to discover improved algorithms,
while RL optimizes the LLM policy based on these discoveries. Our experiments
on three combinatorial optimization tasks - bin packing, traveling salesman,
and the flatpack problem - show that combining RL and evolutionary search
improves discovery efficiency of improved algorithms, showcasing the potential
of RL-enhanced evolutionary strategies to assist computer scientists and
mathematicians for more efficient algorithm design.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 14:14:15 GMT"
},
{
"version": "v2",
"created": "Tue, 8 Apr 2025 21:23:26 GMT"
},
{
"version": "v3",
"created": "Sat, 12 Apr 2025 17:44:16 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Surina",
"Anja",
""
],
[
"Mansouri",
"Amin",
""
],
[
"Quaedvlieg",
"Lars",
""
],
[
"Seddas",
"Amal",
""
],
[
"Viazovska",
"Maryna",
""
],
[
"Abbe",
"Emmanuel",
""
],
[
"Gulcehre",
"Caglar",
""
]
] |
2504.05172 | Guangqiang Li | Guangqiang Li, M. Amine Atoui and Xiangshun Li | Attention-Based Multiscale Temporal Fusion Network for Uncertain-Mode
Fault Diagnosis in Multimode Processes | 31 pages,11 figures | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Fault diagnosis in multimode processes plays a critical role in ensuring the
safe operation of industrial systems across multiple modes. It faces a great
challenge yet to be addressed - that is, the significant distributional
differences among monitoring data from multiple modes make it difficult for the
models to extract shared feature representations related to system health
conditions. In response to this problem, this paper introduces a novel method
called attention-based multiscale temporal fusion network. The multiscale
depthwise convolution and gated recurrent unit are employed to extract
multiscale contextual local features and long-short-term features. Instance
normalization is applied to suppress mode-specific information. Furthermore, a
temporal attention mechanism is designed to focus on critical time points with
higher cross-mode shared information, thereby enhancing the accuracy of fault
diagnosis. The proposed model is applied to Tennessee Eastman process dataset
and three-phase flow facility dataset. The experiments demonstrate that the
proposed model achieves superior diagnostic performance and maintains a small
model size. The source code will be available on GitHub at
https://github.com/GuangqiangLi/AMTFNet.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 15:16:22 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 08:47:52 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Li",
"Guangqiang",
""
],
[
"Atoui",
"M. Amine",
""
],
[
"Li",
"Xiangshun",
""
]
] |
2504.05208 | Alberto Dayan | Alberto Dayan and Daniel Seco | On the cyclic behavior of singular inner functions in Besov and sequence
spaces | null | null | null | null | math.CV math.FA | http://creativecommons.org/licenses/by/4.0/ | We show the existence of singular inner functions that are cyclic in some
Besov-type spaces of analytic functions over the unit disc. Our sufficient
condition is stated only in terms of the modulus of smoothness of the
underlying measure. Such singular inner functions are cyclic also in the space
$\ell^p_A$ of holomorphic functions with coefficients in $\ell^p$. This can
only happen for measures that place no mass on any Beurling-Carleson set.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 15:57:59 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 08:15:49 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Dayan",
"Alberto",
""
],
[
"Seco",
"Daniel",
""
]
] |
2504.05226 | Jungyeul Park | Jungyeul Park | Proposing TAGbank as a Corpus of Tree-Adjoining Grammar Derivations | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | The development of lexicalized grammars, particularly Tree-Adjoining Grammar
(TAG), has significantly advanced our understanding of syntax and semantics in
natural language processing (NLP). While existing syntactic resources like the
Penn Treebank and Universal Dependencies offer extensive annotations for
phrase-structure and dependency parsing, there is a lack of large-scale corpora
grounded in lexicalized grammar formalisms. To address this gap, we introduce
TAGbank, a corpus of TAG derivations automatically extracted from existing
syntactic treebanks. This paper outlines a methodology for mapping
phrase-structure annotations to TAG derivations, leveraging the generative
power of TAG to support parsing, grammar induction, and semantic analysis. Our
approach builds on the work of CCGbank, extending it to incorporate the unique
structural properties of TAG, including its transparent derivation trees and
its ability to capture long-distance dependencies. We also discuss the
challenges involved in the extraction process, including ensuring consistency
across treebank schemes and dealing with language-specific syntactic
idiosyncrasies. Finally, we propose the future extension of TAGbank to include
multilingual corpora, focusing on the Penn Korean and Penn Chinese Treebanks,
to explore the cross-linguistic application of TAG's formalism. By providing a
robust, derivation-based resource, TAGbank aims to support a wide range of
computational tasks and contribute to the theoretical understanding of TAG's
generative capacity.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 16:13:19 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 00:01:05 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Park",
"Jungyeul",
""
]
] |
2504.05257 | Utsav Dewan | Utsav Dewan | Iterated convolution inequalities on $\mathbb{R}^d$ and Riemannian
Symmetric Spaces of non-compact type | Added more results and generalized the result for K-biinvariant
functions to right K-invariant functions | null | null | null | math.FA | http://creativecommons.org/licenses/by/4.0/ | In a recent work (Int Math Res Not 24:18604-18612, 2021),
Carlen-Jauslin-Lieb-Loss studied the convolution inequality $f \ge f*f$ on
$\mathbb{R}^d$ and proved that the real integrable solutions of the above
inequality must be non-negative and satisfy the non-trivial bound
$\int_{\mathbb{R}^d} f \le \frac{1}{2}$. Nakamura-Sawano then generalized their
result to $m$-fold convolution (J Geom Anal 35:68, 2025). In this article, we
replace the monomials by genuine polynomials and study the real-valued
solutions $f \in L^1(\mathbb{R}^d)$ of the iterated convolution inequality
\begin{equation*} f \ge \displaystyle\sum_{n=2}^N a_n \left(*^n f\right) \:,
\end{equation*} where $N \ge 2$ is an integer and for $2 \le n \le N$, $a_n$
are non-negative integers with at least one of them positive. We prove that $f$
must be non-negative and satisfy the non-trivial bound $\int_{\mathbb{R}^d} f
\le t_{\mathcal{Q}}\:$ where $\mathcal{Q}(t):=t-\displaystyle\sum_{n=2}^N
a_n\:t^n$ and $t_{\mathcal{Q}}$ is the unique zero of $\mathcal{Q}'$ in
$(0,\infty)$. We also have an analogue of our result for Riemannian Symmetric
Spaces of non-compact type. Our arguments involve Fourier Analysis and Complex
analysis. We then apply our result to obtain an a priori estimate for solutions
of an integro-differential equation which is related to the physical problem of
the ground state energy of the Bose gas in the classical Euclidean setting.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 16:51:27 GMT"
},
{
"version": "v2",
"created": "Tue, 8 Apr 2025 15:59:33 GMT"
},
{
"version": "v3",
"created": "Sun, 13 Apr 2025 14:53:57 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Dewan",
"Utsav",
""
]
] |
2504.05277 | Federico Ambrosino | Federico Ambrosino, Ingo Runkel and G\'erard M. T. Watts | Non-local charges from perturbed defects via SymTFT in 2d CFT | 66 pages, Mathematica code for the bulk commutation condition in
minimal models is provided in the ancillary files; v2: reference added | null | null | DESY-25-053; Hamburger Beitr. zur Mathematik Nr 989; ZMP-HH/25-6 | hep-th cond-mat.stat-mech math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | We investigate non-local conserved charges in perturbed two-dimensional
conformal field theories from the point of view of the 3d SymTFT of the
unperturbed theory. In the SymTFT we state a simple commutation condition which
results in a pair of compatible bulk and defect perturbations, such that the
perturbed line defects are conserved in the perturbed CFT. In other words, the
perturbed defects are rigidly translation invariant, and such defects form a
monoidal category which extends the topological symmetries. As examples we
study the A-type Virasoro minimal models $M(p,q)$. Our formalism provides
one-parameter families of commuting non-local conserved charges for
perturbations by a primary bulk field with Kac label $(1,2)$, $(1,3)$, or
$(1,5)$, which are the standard integrable perturbations of minimal models. We
find solutions to the commutation condition also for other bulk perturbations,
such as $(1,7)$, and we contrast this with the existence of local conserved
charges. There has been recent interest in the possibility that in certain
cases perturbations by fields such as $(1,7)$ can be integrable, and our
construction provides a new way in which integrability can be found without the
need for local conserved charges.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 17:22:55 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 21:31:33 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ambrosino",
"Federico",
""
],
[
"Runkel",
"Ingo",
""
],
[
"Watts",
"Gérard M. T.",
""
]
] |
2504.05279 | Vitaly Vanchurin | Dmitry Guskov and Vitaly Vanchurin | Covariant Gradient Descent | 12 pages, 2 figures, 2 tables | null | null | null | cs.LG hep-th math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a manifestly covariant formulation of the gradient descent method,
ensuring consistency across arbitrary coordinate systems and general curved
trainable spaces. The optimization dynamics is defined using a covariant force
vector and a covariant metric tensor, both computed from the first and second
statistical moments of the gradients. These moments are estimated through
time-averaging with an exponential weight function, which preserves linear
computational complexity. We show that commonly used optimization methods such
as RMSProp, Adam and AdaBelief correspond to special limits of the covariant
gradient descent (CGD) and demonstrate how these methods can be further
generalized and improved.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 17:25:50 GMT"
},
{
"version": "v2",
"created": "Tue, 8 Apr 2025 20:44:48 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Guskov",
"Dmitry",
""
],
[
"Vanchurin",
"Vitaly",
""
]
] |
2504.05401 | Bringfried Stecklum | Bringfried Stecklum | Disentangling Joint Light Curves Using Photocenter Shifts: A Case Study
Using NEOWISE Data | 3 pages, 1 figure, accepted by RNAAS | null | null | null | astro-ph.IM | http://creativecommons.org/licenses/by/4.0/ | The data collected by the Wide-field Infrared Survey Explorer (WISE, Wright
et al. 2010) and its follow-up Near Earth Object (NEO) mission (NEOWISE,
Mainzer et al. 2011) represent a treasure trove for variability studies.
However, the angular resolution imposed by the primary mirror implies that
close double stars are often unresolved. Then, variability of one or both stars
leads to motion of the image centroid along the connecting line. Knowledge of
the angular separation, derived from higher-resolution imaging which resolves
both components, allows disentangling of the joint light curve into individual
ones. This is illustrated by the case of SPICY 1474 which featured an outburst
several years ago. Removal of the contribution of the nearby companion, which
led to a dilution of the burst strength, revealed that it was ~0.5 mag brighter
than in the joint light curve. A comparison with light curves from unTimely
suggests that utilizing the photocenter shift may lead to more reliable
results.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 18:13:36 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 15:04:37 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Stecklum",
"Bringfried",
""
]
] |
2504.05429 | Sahar Sharifzadeh | L.N. Mohanam, R. Umeda, L. Gu, Y. Song, D.J. Tobias, A.I. Hochbaum, R.
Wu, S. Sharifzadeh | Investigating electron conductivity regimes in the bacterial cytochrome
wire OmcS | null | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | The anaerobic bacterium \textit{Geobacter sulfurreducens} produces
extracellular, electronically conductive cytochrome polymer wires that are
conductive over micron length scales. Structure models from cryo-electron
microscopy data show OmcS wires form a linear chain of hemes along the protein
wire axis, which is proposed as the structural basis supporting their
electronic properties. The geometric arrangement of heme along OmcS wires is
conserved in many multiheme c-type cytochromes and other recently discovered
microbial cytochrome wires. However, the mechanism by which this arrangement of
heme molecules support electron transport through proteins and supramolecular
heme wires is unclear. Here, we investigate the site energies, inter-heme
coupling, and long-range electronic conductivity within OmcS. We introduce an
approach to extract charge carrier site information directly from Kohn-Sham
density functional theory, without employing projector schemes. We show that
site and coupling energies are highly sensitive to changes in inter-heme
geometry and the surrounding electrostatic environment, as intuitively
expected. These parameters serve as inputs for a quantum charge carrier model
that includes decoherence corrections with which we predict a diffusion
coefficient comparable with other organic-based electronic materials. Based on
these simulations, we propose that dynamic disorder, particularly due to
perturbative inter-heme vibrations allow the carrier to overcome trapping due
to the presence of static disorder \textit{via} small frequency-dependent
fluctuations. These studies provide insights into molecular and electronic
determinants of long-range electronic conductivity in microbial cytochrome
wires and highlight design principles for bioinspired, heme-based conductive
materials.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 18:52:11 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Mohanam",
"L. N.",
""
],
[
"Umeda",
"R.",
""
],
[
"Gu",
"L.",
""
],
[
"Song",
"Y.",
""
],
[
"Tobias",
"D. J.",
""
],
[
"Hochbaum",
"A. I.",
""
],
[
"Wu",
"R.",
""
],
[
"Sharifzadeh",
"S.",
""
]
] |
2504.05474 | Ahmad Hoseinpour | Ahmad Hoseinpour, Moslem Zarei and Mehdi Abdi | Photon self-interaction through gravitons and axions | null | null | 10.1140/epjc/s10052-025-14116-x | null | hep-ph quant-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this work, we propose to employ the concept of photon self-interaction for
axion detection. In particular, we derive the interaction Hamiltonian for
photons via axions in a ring cavity. We show that when the incoming photons are
considered in plane-wave basis, the interaction vanishes. However, when the
realistic case of photon wavepackets are assumed, a self-interaction whose
strength is proportional to the size of the wavepacket and the cavity length
exists. Under specific conditions, we find that the axion-mediated interaction
dominates the gravitationally induced self-interaction. We discuss the
implications of this setup for axion detection, focusing on the range of axion
mass, $10^{-10}~\text{eV}<m_a<10^{-4}~\text{eV}$ for which the best accuracy of
coupling constants is constrained in some cases to $g_{a\gamma \gamma}> 9
\times 10^{-12}~ \text{GeV}^{-1}$.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 20:07:30 GMT"
},
{
"version": "v2",
"created": "Wed, 9 Apr 2025 10:37:27 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hoseinpour",
"Ahmad",
""
],
[
"Zarei",
"Moslem",
""
],
[
"Abdi",
"Mehdi",
""
]
] |
2504.05516 | Vasil Yordanov Dr | Vasil Yordanov | Revisiting Lamb Shift Theory through Brownian Motion of the Proton | I am not able to reproduce the Lamb shift for the Deuterium atom and
the 4He+ atom | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | This paper presents a novel theoretical derivation of the Lamb shift in the
hydrogen atom, based solely on fundamental constants and the stochastic
(Brownian) motion of the proton. Unlike conventional quantum electrodynamics
(QED), the proposed approach introduces no experimentally fitted parameters,
offering a fully self-consistent explanation grounded entirely in known
physical quantities.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 21:23:14 GMT"
},
{
"version": "v2",
"created": "Thu, 10 Apr 2025 15:05:01 GMT"
},
{
"version": "v3",
"created": "Sat, 12 Apr 2025 11:16:06 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Yordanov",
"Vasil",
""
]
] |
2504.05522 | Jianling Wang | Jianling Wang, Yifan Liu, Yinghao Sun, Xuejian Ma, Yueqi Wang, He Ma,
Zhengyang Su, Minmin Chen, Mingyan Gao, Onkar Dalal, Ed H. Chi, Lichan Hong,
Ningren Han, Haokai Lu | User Feedback Alignment for LLM-powered Exploration in Large-scale
Recommendation Systems | null | null | null | null | cs.IR | http://creativecommons.org/licenses/by/4.0/ | Exploration, the act of broadening user experiences beyond their established
preferences, is challenging in large-scale recommendation systems due to
feedback loops and limited signals on user exploration patterns. Large Language
Models (LLMs) offer potential by leveraging their world knowledge to recommend
novel content outside these loops. A key challenge is aligning LLMs with user
preferences while preserving their knowledge and reasoning. While using LLMs to
plan for the next novel user interest, this paper introduces a novel approach
combining hierarchical planning with LLM inference-time scaling to improve
recommendation relevancy without compromising novelty. We decouple novelty and
user-alignment, training separate LLMs for each objective. We then scale up the
novelty-focused LLM's inference and select the best-of-n predictions using the
user-aligned LLM. Live experiments demonstrate efficacy, showing significant
gains in both user satisfaction (measured by watch activity and active user
counts) and exploration diversity.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 21:44:12 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 22:16:00 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Wang",
"Jianling",
""
],
[
"Liu",
"Yifan",
""
],
[
"Sun",
"Yinghao",
""
],
[
"Ma",
"Xuejian",
""
],
[
"Wang",
"Yueqi",
""
],
[
"Ma",
"He",
""
],
[
"Su",
"Zhengyang",
""
],
[
"Chen",
"Minmin",
""
],
[
"Gao",
"Mingyan",
""
],
[
"Dalal",
"Onkar",
""
],
[
"Chi",
"Ed H.",
""
],
[
"Hong",
"Lichan",
""
],
[
"Han",
"Ningren",
""
],
[
"Lu",
"Haokai",
""
]
] |
2504.05567 | Jiapeng Zhao | Jiapeng Zhao, Yang Xu, Xiyuan Lu, Eneet Kaur, Michael Kilzer, Ramana
Kompella, Robert W. Boyd, and Reza Nejabati | Scalable MHz-Rate Entanglement Distribution in Low-Latency Quantum
Networks Interconnecting Heterogeneous Quantum Processors | null | null | null | null | quant-ph physics.optics | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Practical distributed quantum computing and error correction require
high-qubit-rate, high-fidelity, and low-reconfiguration-latency quantum
networks between heterogeneous quantum information processors. Unfortunately,
in a quantum network with homogeneous quantum processors, the theoretical
entanglement distribution rate for a single channel is limited to the 100-kHz
level with a millisecond-level reconfiguration latency, which is not sufficient
for error-corrected distributed quantum computing. Here, we propose a quantum
network architecture by introducing the concept of a reconfigurable quantum
interface. In our protocol, through tuning the frequency and temporal mode of
the photonic qubits to dense wavelength division multiplexing (DWDM) channels,
a 4.5 MHz Bell pair distribution rate, with a potential of more than 40 MHz
Bell pair rate, is achieved. Through the use of reconfigurable quantum
interfaces and wavelength-selective switches, a nanosecond network
reconfiguration latency can be demonstrated with low-loss, low-infidelity and
high-dimensional switches. To the best of our knowledge, our architecture is
the first practical solution that can accommodate the entanglement distribution
between heterogeneous quantum nodes with a rate and latency that satisfy most
distributed quantum circuits and error correction requirements. The proposed
architecture is compatible with the industry-standard DWDM infrastructure,
offering a scalable and cost-effective solution for distributed quantum
computing.
| [
{
"version": "v1",
"created": "Mon, 7 Apr 2025 23:47:49 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 22:18:15 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhao",
"Jiapeng",
""
],
[
"Xu",
"Yang",
""
],
[
"Lu",
"Xiyuan",
""
],
[
"Kaur",
"Eneet",
""
],
[
"Kilzer",
"Michael",
""
],
[
"Kompella",
"Ramana",
""
],
[
"Boyd",
"Robert W.",
""
],
[
"Nejabati",
"Reza",
""
]
] |
2504.05579 | Skanda Koppula | Artem Zholus, Carl Doersch, Yi Yang, Skanda Koppula, Viorica
Patraucean, Xu Owen He, Ignacio Rocco, Mehdi S. M. Sajjadi, Sarath Chandar,
Ross Goroshin | TAPNext: Tracking Any Point (TAP) as Next Token Prediction | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-sa/4.0/ | Tracking Any Point (TAP) in a video is a challenging computer vision problem
with many demonstrated applications in robotics, video editing, and 3D
reconstruction. Existing methods for TAP rely heavily on complex
tracking-specific inductive biases and heuristics, limiting their generality
and potential for scaling. To address these challenges, we present TAPNext, a
new approach that casts TAP as sequential masked token decoding. Our model is
causal, tracks in a purely online fashion, and removes tracking-specific
inductive biases. This enables TAPNext to run with minimal latency, and removes
the temporal windowing required by many existing state of art trackers. Despite
its simplicity, TAPNext achieves a new state-of-the-art tracking performance
among both online and offline trackers. Finally, we present evidence that many
widely used tracking heuristics emerge naturally in TAPNext through end-to-end
training. The TAPNext model and code can be found at
https://tap-next.github.io/.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 00:28:42 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 12:17:03 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zholus",
"Artem",
""
],
[
"Doersch",
"Carl",
""
],
[
"Yang",
"Yi",
""
],
[
"Koppula",
"Skanda",
""
],
[
"Patraucean",
"Viorica",
""
],
[
"He",
"Xu Owen",
""
],
[
"Rocco",
"Ignacio",
""
],
[
"Sajjadi",
"Mehdi S. M.",
""
],
[
"Chandar",
"Sarath",
""
],
[
"Goroshin",
"Ross",
""
]
] |
2504.05670 | Renda Han | John Smith, Wenxuan Tu, Junlong Wu, Wenxin Zhang, Jingxin Liu, Haotian
Wang, Jieren Cheng, Huajie Lei, Guangzhen Yao, Lingren Wang, Mengfei Li,
Renda Han, and Yu Li | Dual Boost-Driven Graph-Level Clustering Network | Since I did not obtain the consent of all authors and provided this
version to the arxiv community without authorization, I request to withdraw
the manuscript | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Graph-level clustering remains a pivotal yet formidable challenge in graph
learning. Recently, the integration of deep learning with representation
learning has demonstrated notable advancements, yielding performance
enhancements to a certain degree. However, existing methods suffer from at
least one of the following issues: 1. the original graph structure has noise,
and 2. during feature propagation and pooling processes, noise is gradually
aggregated into the graph-level embeddings through information propagation.
Consequently, these two limitations mask clustering-friendly information,
leading to suboptimal graph-level clustering performance. To this end, we
propose a novel Dual Boost-Driven Graph-Level Clustering Network (DBGCN) to
alternately promote graph-level clustering and filtering out interference
information in a unified framework. Specifically, in the pooling step, we
evaluate the contribution of features at the global and optimize them using a
learnable transformation matrix to obtain high-quality graph-level
representation, such that the model's reasoning capability can be improved.
Moreover, to enable reliable graph-level clustering, we first identify and
suppress information detrimental to clustering by evaluating similarities
between graph-level representations, providing more accurate guidance for
multi-view fusion. Extensive experiments demonstrated that DBGCN outperforms
the state-of-the-art graph-level clustering methods on six benchmark datasets.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 04:32:46 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 05:49:45 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Smith",
"John",
""
],
[
"Tu",
"Wenxuan",
""
],
[
"Wu",
"Junlong",
""
],
[
"Zhang",
"Wenxin",
""
],
[
"Liu",
"Jingxin",
""
],
[
"Wang",
"Haotian",
""
],
[
"Cheng",
"Jieren",
""
],
[
"Lei",
"Huajie",
""
],
[
"Yao",
"Guangzhen",
""
],
[
"Wang",
"Lingren",
""
],
[
"Li",
"Mengfei",
""
],
[
"Han",
"Renda",
""
],
[
"Li",
"Yu",
""
]
] |
2504.05835 | Mian Zhu | Alexander Ganz, Chunshan Lin, Mian Zhu | Scalar-assisted magnetogenesis during the radiation-dominated epoch | null | null | null | null | hep-ph astro-ph.CO gr-qc | http://creativecommons.org/licenses/by/4.0/ | We propose a novel mechanism to generate primordial magnetic fields (PMFs)
strong enough to explain the observed cosmic magnetic fields. We employ a
scalar field charged under U(1) gauge symmetry with a non-trivial VEV to
provide an effective mass term to the EM field and thus break its conformal
invariance. The primordial magneto-genesis takes place in the radiation
dominated (RD) epoch, after the electroweak symmetry breaking (EWSB) phase. As
a result, our mechanism is naturally free from the over-production of electric
fields due to high conductivity in the RD epoch, and the baryon isocurvature
problem which takes place only if magneto-genesis happens before the ESWB
phase. In addition, we find that a significant amount of PMFs can be generated
when the scalar field experiences a tachyonic phase. In this case, the scalar
field is light and weakly coupled and has negligible energy density compared to
the cold dark matter, hence the strong coupling problem and the back-reaction
problem are also absent. Therefore, our model is free from the above-mentioned
problems that frequently appear in other primordial magneto-genesis scenarios.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 09:18:30 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 09:25:41 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ganz",
"Alexander",
""
],
[
"Lin",
"Chunshan",
""
],
[
"Zhu",
"Mian",
""
]
] |
2504.05946 | Ruixiang Wu | Ruixiang Wu and Jiahao Ai and Tongxin Li | InstructMPC: A Human-LLM-in-the-Loop Framework for Context-Aware Control | null | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | Model Predictive Control (MPC) is a powerful control strategy widely utilized
in domains like energy management, building control, and autonomous systems.
However, its effectiveness in real-world settings is challenged by the need to
incorporate context-specific predictions and expert instructions, which
traditional MPC often neglects. We propose InstructMPC, a novel framework that
addresses this gap by integrating real-time human instructions through a Large
Language Model (LLM) to produce context-aware predictions for MPC. Our method
employs a Language-to-Distribution (L2D) module to translate contextual
information into predictive disturbance trajectories, which are then
incorporated into the MPC optimization. Unlike existing context-aware and
language-based MPC models, InstructMPC enables dynamic human-LLM interaction
and fine-tunes the L2D module in a closed loop with theoretical performance
guarantees, achieving a regret bound of $O(\sqrt{T\log T})$ for linear dynamics
when optimized via advanced fine-tuning methods such as Direct Preference
Optimization (DPO) using a tailored loss function.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 11:59:00 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 12:28:02 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Wu",
"Ruixiang",
""
],
[
"Ai",
"Jiahao",
""
],
[
"Li",
"Tongxin",
""
]
] |
2504.06006 | Roman Kochnev | Roman Kochnev, Arash Torabi Goodarzi, Zofia Antonina Bentyn, Dmitry
Ignatov, Radu Timofte | Optuna vs Code Llama: Are LLMs a New Paradigm for Hyperparameter Tuning? | null | null | null | null | cs.LG cs.AI cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Optimal hyperparameter selection is critical for maximizing neural network
performance, especially as models grow in complexity. This work investigates
the viability of leveraging large language models (LLMs) for hyperparameter
optimization by fine-tuning a parameter-efficient version of Code Llama using
LoRA. The adapted LLM is capable of generating accurate and efficient
hyperparameter recommendations tailored to diverse neural network
architectures. Unlike traditional approaches such as Optuna, which rely on
computationally intensive trial-and-error procedures, our method achieves
competitive or superior results in terms of Root Mean Square Error (RMSE) while
significantly reducing computational overhead. Our findings demonstrate that
LLM-based optimization not only matches the performance of state-of-the-art
techniques like Tree-structured Parzen Estimators (TPE) but also substantially
accelerates the tuning process. This positions LLMs as a promising alternative
for rapid experimentation, particularly in resource-constrained environments
such as edge devices and mobile platforms, where computational efficiency is
essential. In addition to improved efficiency, the method offers time savings
and consistent performance across various tasks, highlighting its robustness
and generalizability. All generated hyperparameters are included in the LEMUR
Neural Network (NN) Dataset, which is publicly available and serves as an
open-source benchmark for hyperparameter optimization research.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 13:15:47 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 20:43:00 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Kochnev",
"Roman",
""
],
[
"Goodarzi",
"Arash Torabi",
""
],
[
"Bentyn",
"Zofia Antonina",
""
],
[
"Ignatov",
"Dmitry",
""
],
[
"Timofte",
"Radu",
""
]
] |
2504.06065 | Laura Duffy | Laura Duffy, Michael Eracleous, John J. Ruan, Qian Yang, Jessie C.
Runnoe | Sleeping Giants Arise: Monitoring the Return of Three Changing-Look
Quasars to Their High States | 24 pages, 10 figures, 5 tables, Submitted to ApJ | null | null | null | astro-ph.GA astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Changing-look quasars challenge many models of the quasar central engine.
Their extreme variability in both the continuum and broad emission-line fluxes
on timescales on the order of years is difficult to explain. To investigate the
cause of the observed transitions, we present new contemporaneous optical and
X-ray observations of three faded changing-look quasars as they return to the
high state. Two of these three faded changing-look quasars remained in a
quiescent state for more than ten years before returning to a new high state.
We find that before, during, and after transition, the spectral energy
distributions of all three follow predictions for quasars based on X-ray binary
outbursts, suggesting that the mechanism for the change is likely a changing
accretion rate causing changes in the accretion flow structure. We also find
that, in two of the three cases, the transition between the initial high and
low state and the transition between the low and new high state took nearly
identical amounts of time, on the order of hundreds of days. This transition
timescale is useful for testing theoretical models that attempt to explain the
short time scale for the state transition.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 14:08:39 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 18:26:35 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Duffy",
"Laura",
""
],
[
"Eracleous",
"Michael",
""
],
[
"Ruan",
"John J.",
""
],
[
"Yang",
"Qian",
""
],
[
"Runnoe",
"Jessie C.",
""
]
] |
2504.06106 | Vincenzo Petrone | Vincenzo Petrone, Enrico Ferrentino, Pasquale Chiacchio | A ROS2-based software library for inverse dynamics computation | 6 pages, 8 figures | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Inverse dynamics computation is a critical component in robot control,
planning and simulation, enabling the calculation of joint torques required to
achieve a desired motion. This paper presents a ROS2-based software library
designed to solve the inverse dynamics problem for robotic systems. The library
is built around an abstract class with three concrete implementations: one for
simulated robots and two for real UR10 and Franka robots. This contribution
aims to provide a flexible, extensible, robot-agnostic solution to inverse
dynamics, suitable for both simulation and real-world scenarios involving
planning and control applications. The related software is available at
https://github.com/unisa-acg/inverse-dynamics-solver/tree/rap.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 14:50:17 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 10:46:00 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Petrone",
"Vincenzo",
""
],
[
"Ferrentino",
"Enrico",
""
],
[
"Chiacchio",
"Pasquale",
""
]
] |
2504.06112 | Alfonso De Miguel-Arribas | Chengbin Sun, Alfonso de Miguel-Arribas, Chaoqian Wang, Haoxiang Xia,
Yamir Moreno | Co-evolution of cooperation and resource allocation in the advantageous
environment-based spatial multi-game using adaptive control | 30 pages, 18 figures | null | null | null | physics.soc-ph | http://creativecommons.org/licenses/by/4.0/ | In real-life complex systems, individuals often encounter multiple social
dilemmas that cannot be effectively captured using a single-game model.
Furthermore, the environment and limited resources both play a crucial role in
shaping individuals' decision-making behaviors. In this study, we employ an
adaptive control mechanism by which agents may benefit from their environment,
thus redefining their individual fitness. Under this setting, a detailed
examination of the co-evolution of individual strategies and resource
allocation is carried. Through extensive simulations, we find that the
advantageous environment mechanism not only significantly increases the
proportion of cooperators in the system but also influences the resource
distribution among individuals. Additionally, limited resources reinforce
cooperative behaviors within the system while shaping the evolutionary dynamics
and strategic interactions across different dilemmas. Once the system reaches
equilibrium, resource distribution becomes highly imbalanced. To promote fairer
resource allocation, we introduce a minimum resource guarantee mechanism. Our
results show that this mechanism not only reduces disparities in resource
distribution across the entire system and among individuals in different
dilemmas but also significantly enhances cooperative behavior in higher
resource intervals. Finally, to assess the robustness of our model, we further
examine the influence of the advantageous environment on system-wide
cooperation in small-world and random graph network models.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 15:00:31 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 19:33:17 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Sun",
"Chengbin",
""
],
[
"de Miguel-Arribas",
"Alfonso",
""
],
[
"Wang",
"Chaoqian",
""
],
[
"Xia",
"Haoxiang",
""
],
[
"Moreno",
"Yamir",
""
]
] |
2504.06136 | Movina Moses | Movina Moses, Mohab Elkaref, James Barry, Shinnosuke Tanaka, Vishnudev
Kuruvanthodi, Nathan Herr, Campbell D Watson, Geeth De Mel | QGen Studio: An Adaptive Question-Answer Generation, Training and
Evaluation Platform | null | null | 10.1609/aaai.v39i28.35362 | null | cs.CL cs.AI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We present QGen Studio: an adaptive question-answer generation, training, and
evaluation platform. QGen Studio enables users to leverage large language
models (LLMs) to create custom question-answer datasets and fine-tune models on
this synthetic data. It features a dataset viewer and model explorer to
streamline this process. The dataset viewer provides key metrics and visualizes
the context from which the QA pairs are generated, offering insights into data
quality. The model explorer supports model comparison, allowing users to
contrast the performance of their trained LLMs against other models, supporting
performance benchmarking and refinement. QGen Studio delivers an interactive,
end-to-end solution for generating QA datasets and training scalable,
domain-adaptable models. The studio will be open-sourced soon, allowing users
to deploy it locally.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 15:32:09 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Moses",
"Movina",
""
],
[
"Elkaref",
"Mohab",
""
],
[
"Barry",
"James",
""
],
[
"Tanaka",
"Shinnosuke",
""
],
[
"Kuruvanthodi",
"Vishnudev",
""
],
[
"Herr",
"Nathan",
""
],
[
"Watson",
"Campbell D",
""
],
[
"De Mel",
"Geeth",
""
]
] |
2504.06141 | Alexander Bukharin | Alexander Bukharin, Haifeng Qian, Shengyang Sun, Adithya
Renduchintala, Soumye Singhal, Zhilin Wang, Oleksii Kuchaiev, Olivier
Delalleau, Tuo Zhao | Adversarial Training of Reward Models | 16 pages, 7 figures | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Reward modeling has emerged as a promising approach for the scalable
alignment of language models. However, contemporary reward models (RMs) often
lack robustness, awarding high rewards to low-quality, out-of-distribution
(OOD) samples. This can lead to reward hacking, where policies exploit
unintended shortcuts to maximize rewards, undermining alignment. To address
this challenge, we introduce Adv-RM, a novel adversarial training framework
that automatically identifies adversarial examples -- responses that receive
high rewards from the target RM but are OOD and of low quality. By leveraging
reinforcement learning, Adv-RM trains a policy to generate adversarial examples
that reliably expose vulnerabilities in large state-of-the-art reward models
such as Nemotron 340B RM. Incorporating these adversarial examples into the
reward training process improves the robustness of RMs, mitigating reward
hacking and enhancing downstream performance in RLHF. We demonstrate that
Adv-RM significantly outperforms conventional RM training, increasing stability
and enabling more effective RLHF training in both synthetic and real-data
settings.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 15:38:25 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 18:28:40 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Bukharin",
"Alexander",
""
],
[
"Qian",
"Haifeng",
""
],
[
"Sun",
"Shengyang",
""
],
[
"Renduchintala",
"Adithya",
""
],
[
"Singhal",
"Soumye",
""
],
[
"Wang",
"Zhilin",
""
],
[
"Kuchaiev",
"Oleksii",
""
],
[
"Delalleau",
"Olivier",
""
],
[
"Zhao",
"Tuo",
""
]
] |
2504.06160 | Rijul Magu | Rijul Magu, Arka Dutta, Sean Kim, Ashiqur R. KhudaBukhsh, Munmun De
Choudhury | Navigating the Rabbit Hole: Emergent Biases in LLM-Generated Attack
Narratives Targeting Mental Health Groups | null | null | null | null | cs.CL cs.AI cs.CY cs.LG cs.SI | http://creativecommons.org/licenses/by/4.0/ | Large Language Models (LLMs) have been shown to demonstrate imbalanced biases
against certain groups. However, the study of unprovoked targeted attacks by
LLMs towards at-risk populations remains underexplored. Our paper presents
three novel contributions: (1) the explicit evaluation of LLM-generated attacks
on highly vulnerable mental health groups; (2) a network-based framework to
study the propagation of relative biases; and (3) an assessment of the relative
degree of stigmatization that emerges from these attacks. Our analysis of a
recently released large-scale bias audit dataset reveals that mental health
entities occupy central positions within attack narrative networks, as revealed
by a significantly higher mean centrality of closeness (p-value = 4.06e-10) and
dense clustering (Gini coefficient = 0.7). Drawing from sociological
foundations of stigmatization theory, our stigmatization analysis indicates
increased labeling components for mental health disorder-related targets
relative to initial targets in generation chains. Taken together, these
insights shed light on the structural predilections of large language models to
heighten harmful discourse and highlight the need for suitable approaches for
mitigation.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 15:56:57 GMT"
},
{
"version": "v2",
"created": "Wed, 9 Apr 2025 04:24:38 GMT"
},
{
"version": "v3",
"created": "Fri, 11 Apr 2025 20:13:11 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Magu",
"Rijul",
""
],
[
"Dutta",
"Arka",
""
],
[
"Kim",
"Sean",
""
],
[
"KhudaBukhsh",
"Ashiqur R.",
""
],
[
"De Choudhury",
"Munmun",
""
]
] |
2504.06204 | Adriane Consuelo Leal Auccaise Leal | Adriane Consuelo Leal Auccaise and Antonio S\'ergio Magalh\~aes de
Castro | Spin Squeezing via One-Axis Twisting in a Quadrupolar NMR system under
relaxation effects | null | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study investigates spin squeezed states in nuclear magnetic resonance
(NMR) quadrupolar systems with spins $I=3/2$ and $I=7/2$ at room temperature,
taking into account the effects of relaxation on the dynamics. The origin of
spin squeezing is attributed to the interaction between the nuclear quadrupole
moment and the electric field gradients in the molecular environment. The
formal description of the nonlinear operators responsible for spin squeezing is
achieved using the spin angular momentum representation via the one-axis
twisting mechanism. This approach provides a framework for quantum control and
metrology over the spin squeezing process while accounting for the influence of
relaxation phenomena. Non-Cartesian angular momentum operators are proposed,
with their variances products catching the quantum effects during the dynamics.
An upper bound for the squeezing parameter and the Heisenberg uncertainty at
thermal equilibrium are also predicted for any spin quantum number.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 16:47:41 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 01:38:37 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Auccaise",
"Adriane Consuelo Leal",
""
],
[
"de Castro",
"Antonio Sérgio Magalhães",
""
]
] |
2504.06280 | E M Hasantha Ekanayake | E.M.H.E.B. Ekanayake and N. Shukla | Different Paths, Same Destination: Designing New Physics-Inspired
Dynamical Systems with Engineered Stability to Minimize the Ising Hamiltonian | 11 pages, 3 figures | null | null | null | cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Oscillator Ising machines (OIMs) represent an exemplar case of using
physics-inspired non-linear dynamical systems to solve computationally
challenging combinatorial optimization problems (COPs). The computational
performance of such systems is highly sensitive to the underlying dynamical
properties, the topology of the input graph, and their relative compatibility.
In this work, we explore the concept of designing different dynamical systems
that minimize the same objective function but exhibit drastically different
dynamical properties. Our goal is to leverage this diversification in dynamics
to reduce the sensitivity of the computational performance to the underlying
graph, and subsequently, enhance the overall effectiveness of such
physics-based computational methods. To this end, we introduce a novel
dynamical system, the Dynamical Ising Machine (DIM), which, like the OIM,
minimizes the Ising Hamiltonian but offers significantly different dynamical
properties. We analyze the characteristic properties of the DIM and compare
them with those of the OIM. We also show that the relative performance of each
model is dependent on the input graph. Our work illustrates that using multiple
dynamical systems with varying properties to solve the same COP enables an
effective method that is less sensitive to the input graph, while producing
robust solutions.
| [
{
"version": "v1",
"created": "Fri, 21 Mar 2025 21:24:22 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 22:45:48 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ekanayake",
"E. M. H. E. B.",
""
],
[
"Shukla",
"N.",
""
]
] |
2504.06310 | Changqing Shen | Changqing Shen, BingZhou Xu, Xiaojian Zhang, Sijie Yan, Han Ding | Conformal Slit Mapping Based Spiral Tool Trajectory Planning for
Ball-end Milling on Complex Freeform Surfaces | The revised manuscript has improved the quality of the figures | null | null | null | cs.GR | http://creativecommons.org/licenses/by/4.0/ | This study presents a spiral-based complete coverage strategy for ball-end
milling on freeform surfaces, utilizing conformal slit mapping to generate
milling trajectories that are more compact, smoother, and evenly distributed
when machining 2D cavities with islands. This approach, an upgrade from
traditional methods, extends the original algorithm to effectively address 3D
perforated surface milling. Unlike conventional algorithms, the method embeds a
continuous spiral trajectory within perforated surfaces without requiring
cellular decomposition or additional boundaries. The proposed method addresses
three primary challenges, including modifying conformal slit mapping for mesh
surfaces, maintaining uniform scallop height between adjacent spiral
trajectories, and optimizing the mapped origin point to ensure uniform scallop
height distribution. To overcome these challenges, surface flattening
techniques are incorporated into the original approach to accommodate mesh
surfaces effectively. Tool path spacing is then optimized using a binary search
strategy to regulate scallop height. A functional energy metric associated with
scallop height uniformity is introduced for rapid evaluation of points mapped
to the origin, with the minimum functional energy determined through
perturbation techniques. The optimal placement of this point is identified
using a modified gradient descent approach applied to the energy function.
Validation on intricate surfaces, including low-quality and high-genus meshes,
verifies the robustness of the algorithm. Surface milling experiments comparing
this method with conventional techniques indicate a 15.63% improvement in
scallop height uniformity while reducing machining time, average spindle
impact, and spindle impact variance by up to 7.36%, 27.79%, and 55.98%,
respectively.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 02:12:13 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 02:29:56 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Shen",
"Changqing",
""
],
[
"Xu",
"BingZhou",
""
],
[
"Zhang",
"Xiaojian",
""
],
[
"Yan",
"Sijie",
""
],
[
"Ding",
"Han",
""
]
] |
2504.06313 | Abdulkareem Alsudais | Abdulkareem Alsudais | Text-to-Image Models and Their Representation of People from Different
Nationalities Engaging in Activities | null | null | null | null | cs.CV cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The primary objective of this paper is to investigate how a popular
Text-to-Image (T2I) model represents people from 208 different nationalities
when prompted to generate images of individuals performing typical activities.
Two scenarios were developed, and images were generated based on input prompts
that specified nationalities. The results show that in one scenario, the
majority of images, and in the other, a substantial portion, depict individuals
wearing traditional attire. This suggests that the model emphasizes such
characteristics even when they are impractical for the given activity. A
statistically significant relationship was observed between this representation
pattern and the regions associated with the specified countries. This indicates
that the issue disproportionately affects certain areas, particularly the
Middle East & North Africa and Sub-Saharan Africa. A notable association with
income groups was also found. CLIP was used to measure alignment scores between
generated images and various prompts and captions. The findings indicate
statistically significant higher scores for images featuring individuals in
traditional attire in one scenario. The study also examined revised prompts
(additional contextual information automatically added to the original input
prompts) to assess their potential influence on how individuals are represented
in the generated images, finding that the word "traditional" was commonly added
to revised prompts. These findings provide valuable insights into how T2I
models represent individuals from various countries and highlight potential
areas for improvement in future models.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 05:37:06 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 04:37:55 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Alsudais",
"Abdulkareem",
""
]
] |
2504.06334 | Richard Pan | Richard Pan, Katherine A. Suess, Danilo Marchesini, Bingjie Wang, Joel
Leja, Sam E. Cutler, Katherine E. Whitaker, Rachel Bezanson, Sedona H. Price,
Lukas J. Furtak, John R. Weaver, Ivo Labb\'e, Gabriel Brammer, Yunchong
Zhang, Pratika Dayal, Robert Feldmann, Jenny E. Greene, Tim B. Miller, Ikki
Mitsuhashi, Themiya Nanayakkara, Erica J. Nelson, David J. Setton, and Adi
Zitrin | UNCOVER/MegaScience: No Evidence of Environmental Quenching in a
z$\sim$2.6 Proto-cluster | 12 pages, 6 figures, Submitted to ApJL | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Environmental quenching -- where interactions with other galaxies and/or the
intra-cluster medium (ICM) suppress star formation in low-mass galaxies -- has
long been proposed as the primary driver to establish the red sequence for
low-mass galaxies within clusters at low redshift ($z<1$). However, we still do
not know whether these environmental quenching mechanisms are also active at
higher redshifts in proto-cluster environments that have yet to fully
virialize. In large part, this regime has remained unexplored due to
observational limitations; however, the James Webb Space Telescope has recently
opened a new window into the role of environmental quenching on low-mass
(log(M$_{\star}$/M$_{\odot}$)$<$9.5) galaxies at cosmic noon ($2 < z < 3$).
Here, we use data from the JWST UNCOVER and MegaScience programs to directly
probe the role of environmental quenching on low-mass galaxies in a
newly-discovered $z\approx 2.6$ overdensity. Leveraging the deep imaging and R
$\sim$ 15 spectrophotometry enabled by these JWST/NIRCam data, we analyze the
stellar populations and inferred star formation histories (SFHs) of 20 low-mass
(8.5$<$log(M$_{\star}$/M$_{\odot}$)$\leq$9.0) quiescent galaxies in the
overdense environment and compare to a similar sample of 18 such galaxies in
the field. The SFHs of quiescent galaxies in the proto-cluster and field across
the entire probed stellar mass regime
(8.5$<$log(M$_{\star}$/M$_{\odot}$)$\leq$11.0) are indistinguishable,
demonstrating that the environment at cosmic noon is not yet accelerating
quenching compared to the field. This is consistent with expectations that
proto-clusters at $z>2$ have yet to virialize and develop a dense enough
environment to efficiently quench low-mass galaxies.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 18:00:00 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 15:07:15 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Pan",
"Richard",
""
],
[
"Suess",
"Katherine A.",
""
],
[
"Marchesini",
"Danilo",
""
],
[
"Wang",
"Bingjie",
""
],
[
"Leja",
"Joel",
""
],
[
"Cutler",
"Sam E.",
""
],
[
"Whitaker",
"Katherine E.",
""
],
[
"Bezanson",
"Rachel",
""
],
[
"Price",
"Sedona H.",
""
],
[
"Furtak",
"Lukas J.",
""
],
[
"Weaver",
"John R.",
""
],
[
"Labbé",
"Ivo",
""
],
[
"Brammer",
"Gabriel",
""
],
[
"Zhang",
"Yunchong",
""
],
[
"Dayal",
"Pratika",
""
],
[
"Feldmann",
"Robert",
""
],
[
"Greene",
"Jenny E.",
""
],
[
"Miller",
"Tim B.",
""
],
[
"Mitsuhashi",
"Ikki",
""
],
[
"Nanayakkara",
"Themiya",
""
],
[
"Nelson",
"Erica J.",
""
],
[
"Setton",
"David J.",
""
],
[
"Zitrin",
"Adi",
""
]
] |
2504.06339 | Amir H. Khoram | Amir H. Khoram, Bianca Poggianti, Alessia Moretti, Benedetta Vulcani,
Mario Radovich, Ariel Werle, Marco Gullieuszik, Amirnezam Amiri, Sirio Belli,
Letizia Bugiani, Neven Tomicic, Giorgia Peluso, Eric Giunchi, Johan Richard | Stripped and Enriched: The Role of Ram-Pressure in Shaping Chemical
Enrichment of Galaxies at Intermediate Redshift | Accepted for publication in MNRAS | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | The chemical evolution of galaxies is shaped by their star formation
histories and the exchange of gas with their environments. Metallicity provides
key insights into these processes, reflecting the interplay between star
formation and gas flows. A fundamental aspect of this evolution is the
mass-metallicity relation, which captures the strong correlation between a
galaxy stellar mass ($M_\star$) and its gas-phase oxygen abundance. In this
study, we use MUSE observations to analyze star-forming disc galaxies in 12
clusters within the redshift range $0.3 < z < 0.5$. Galaxies were classified
into three groups: ram-pressure stripping (RPS), control cluster, and control
field. For the first time, we investigate the impact of RPS on gas-phase
metallicities across a wide mass range of galaxies at intermediate redshift,
comparing RPS galaxies to counterparts in both cluster and field environments.
By analyzing the integrated flux within galactic disks, our results reveal
that, on average, RPS induces a metallicity enhancement of 0.2 dex over
non-stripped galaxies. Contrary to the prevailing view that cluster membership
alone drives metallicity enrichment, we find that control cluster galaxies
exhibit metallicities comparable to field galaxies at a given $M_\star$, with
only RPS galaxies displaying significantly higher metal content, highlighting
the unique role of RPS in shaping the chemical properties of galaxies. These
differences become more pronounced at lower $M_\star$, indicating that
environmental influences play a more critical role in shaping the chemical
evolution of lower-mass galaxies. Our findings suggest that both enhanced star
formation rates and suppressed gas inflows -- consequences of ram pressure
stripping -- drive the elevated metallicity observed in RPS galaxies.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 18:00:01 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 12:40:21 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Khoram",
"Amir H.",
""
],
[
"Poggianti",
"Bianca",
""
],
[
"Moretti",
"Alessia",
""
],
[
"Vulcani",
"Benedetta",
""
],
[
"Radovich",
"Mario",
""
],
[
"Werle",
"Ariel",
""
],
[
"Gullieuszik",
"Marco",
""
],
[
"Amiri",
"Amirnezam",
""
],
[
"Belli",
"Sirio",
""
],
[
"Bugiani",
"Letizia",
""
],
[
"Tomicic",
"Neven",
""
],
[
"Peluso",
"Giorgia",
""
],
[
"Giunchi",
"Eric",
""
],
[
"Richard",
"Johan",
""
]
] |
2504.06371 | Yibo Shi | Yibo Shi and Cristian R. Rojas | Efficient Simulation of Singularly Perturbed Systems Using a Stabilized
Multirate Explicit Scheme | Accepted by ECC 2025 | null | null | null | math.NA cs.NA cs.SY eess.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Singularly perturbed systems (SPSs) are prevalent in engineering
applications, where numerically solving their initial value problems (IVPs) is
challenging due to stiffness arising from multiple time scales. Classical
explicit methods require impractically small time steps for stability, while
implicit methods developed for SPSs are computationally intensive and less
efficient for strongly nonlinear systems. This paper introduces a Stabilized
Multirate Explicit Scheme (SMES) that stabilizes classical explicit methods
without the need for small time steps or implicit formulations. By employing a
multirate approach with variable time steps, SMES allows the fast dynamics to
rapidly converge to their equilibrium manifold while slow dynamics evolve with
larger steps. Analysis shows that SMES achieves numerical stability with
significantly reduced computational effort and controlled error. Its
effectiveness is illustrated with a numerical example.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 18:40:53 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 19:39:57 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Shi",
"Yibo",
""
],
[
"Rojas",
"Cristian R.",
""
]
] |
2504.06375 | Quentin Buat | PIONEER Collaboration: A. Adelmann, W. Altmannshofer, S. Ban, O.
Beesley, A. Bolotnikov, T. Brunner, D. Bryman, Q. Buat, L. Caminada, J.
Carlton, S. Chen, M. Chiu, V. Cirigliano, S. Corrodi, A. Crivellin, S.
Cuen-Rochin, J. Datta, B. Davis-Purcell, A. Deshpande, A. Di Canto, A.
Ebrahimi, P. Fisher, S. Foster, K. Frahm, L. Gerritzen, G. Giacomini, L.
Gibbons, C. Glaser, T. Gorringe, C. Hamilton, S. Heinekamp, D. Hertzog, S.
Hochrein, M. Hoferichter, S. Ito, T. Iwamoto, P. Kammel, E. Klemets, L.
Kurchaninov, K. Labe, U. Langenegger, Y. Li, C. Malbrunot, A. Matsushita, S.
M. Mazza, S. Mehrotra, S.Mihara, R. Mischke, A. Molnar, T. Mori, T. Numao, W.
Ootani, J. Ott, K. Pachal, D. Pocanic, X. Qian, D. Ries, R. Roehnelt, T.
Rohe, T. Rostomyan, B. Schumm, P. Schwendimann, A. Seiden, A. Sher, R.
Shrock, A. Soter, T. Sullivan, E. Swanson, V. Tishchenko, A. Tricoli, T.
Tsang, Y. Uchiyama, B. Velghe, M. Worcester, E. Worcester, J. Yang, Y. Zhang,
C. Zhang | European Strategy for Particle Physics Update -- PIONEER: a next
generation rare pion decay experiment | Input to the 2024-2026 Update of the of the European Strategy for
Particle Physics | null | null | null | hep-ex nucl-ex physics.ins-det | http://creativecommons.org/licenses/by/4.0/ | PIONEER is a rapidly developing effort aimed to perform a pristine test of
lepton flavour universality (LFU) and of the unitarity of the first row of the
CKM matrix by significantly improving the measurements of rare decays of the
charged pion. In Phase I, PIONEER aims to measure the charged-pion branching
ratio to electrons vs.\ muons $R_{e/\mu}$ to 1 part in $10^4$, improving the
current experimental result $R_{e/\mu}\,\text{(exp)} =1.2327(23)\times10^{-4}$
by a factor of 15. This precision on $R_{e/\mu}$ will match the theoretical
accuracy of the SM prediction allowing for a test of LFU at an unprecedented
level, probing non-SM explanations of LFU violation through sensitivity to
quantum effects of new particles up to the PeV mass scale. Phase II and III
will aim to improve the experimental precision of the branching ratio of pion
beta decay, $\pi^+\to \pi^0 e^+ \nu (\gamma)$, currently at
$1.036(6)\times10^{-8}$, by a factor of three and six, respectively. The
improved measurements will be used to extract $V_{ud}$ in a theoretically
pristine manner. The ultimate precision of $V_{ud}$ is expected to reach the
0.05\,\% level, allowing for a stringent test of CKM unitarity. The PIONEER
experiment will also improve the experimental limits by an order of magnitude
or more on a host of exotic decays that probe the effects of heavy neutrinos
and dark sector physics. This input to the 2026 update of the European Strategy
for Particle Physics Strategy describes the physics motivation and the
conceptual design of the PIONEER experiment, and is prepared based on the
PIONEER proposal submitted to and approved with high priority by the PSI
program advisory committee (PAC). Using intense pion beams, and
state-of-the-art instrumentation and computational resources, the PIONEER
experiment is aiming to begin data taking by the end of this decade.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 18:46:11 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 17:54:54 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"PIONEER Collaboration",
"",
""
],
[
"Adelmann",
"A.",
""
],
[
"Altmannshofer",
"W.",
""
],
[
"Ban",
"S.",
""
],
[
"Beesley",
"O.",
""
],
[
"Bolotnikov",
"A.",
""
],
[
"Brunner",
"T.",
""
],
[
"Bryman",
"D.",
""
],
[
"Buat",
"Q.",
""
],
[
"Caminada",
"L.",
""
],
[
"Carlton",
"J.",
""
],
[
"Chen",
"S.",
""
],
[
"Chiu",
"M.",
""
],
[
"Cirigliano",
"V.",
""
],
[
"Corrodi",
"S.",
""
],
[
"Crivellin",
"A.",
""
],
[
"Cuen-Rochin",
"S.",
""
],
[
"Datta",
"J.",
""
],
[
"Davis-Purcell",
"B.",
""
],
[
"Deshpande",
"A.",
""
],
[
"Di Canto",
"A.",
""
],
[
"Ebrahimi",
"A.",
""
],
[
"Fisher",
"P.",
""
],
[
"Foster",
"S.",
""
],
[
"Frahm",
"K.",
""
],
[
"Gerritzen",
"L.",
""
],
[
"Giacomini",
"G.",
""
],
[
"Gibbons",
"L.",
""
],
[
"Glaser",
"C.",
""
],
[
"Gorringe",
"T.",
""
],
[
"Hamilton",
"C.",
""
],
[
"Heinekamp",
"S.",
""
],
[
"Hertzog",
"D.",
""
],
[
"Hochrein",
"S.",
""
],
[
"Hoferichter",
"M.",
""
],
[
"Ito",
"S.",
""
],
[
"Iwamoto",
"T.",
""
],
[
"Kammel",
"P.",
""
],
[
"Klemets",
"E.",
""
],
[
"Kurchaninov",
"L.",
""
],
[
"Labe",
"K.",
""
],
[
"Langenegger",
"U.",
""
],
[
"Li",
"Y.",
""
],
[
"Malbrunot",
"C.",
""
],
[
"Matsushita",
"A.",
""
],
[
"Mazza",
"S. M.",
""
],
[
"Mehrotra",
"S.",
""
],
[
"Mihara",
"S.",
""
],
[
"Mischke",
"R.",
""
],
[
"Molnar",
"A.",
""
],
[
"Mori",
"T.",
""
],
[
"Numao",
"T.",
""
],
[
"Ootani",
"W.",
""
],
[
"Ott",
"J.",
""
],
[
"Pachal",
"K.",
""
],
[
"Pocanic",
"D.",
""
],
[
"Qian",
"X.",
""
],
[
"Ries",
"D.",
""
],
[
"Roehnelt",
"R.",
""
],
[
"Rohe",
"T.",
""
],
[
"Rostomyan",
"T.",
""
],
[
"Schumm",
"B.",
""
],
[
"Schwendimann",
"P.",
""
],
[
"Seiden",
"A.",
""
],
[
"Sher",
"A.",
""
],
[
"Shrock",
"R.",
""
],
[
"Soter",
"A.",
""
],
[
"Sullivan",
"T.",
""
],
[
"Swanson",
"E.",
""
],
[
"Tishchenko",
"V.",
""
],
[
"Tricoli",
"A.",
""
],
[
"Tsang",
"T.",
""
],
[
"Uchiyama",
"Y.",
""
],
[
"Velghe",
"B.",
""
],
[
"Worcester",
"M.",
""
],
[
"Worcester",
"E.",
""
],
[
"Yang",
"J.",
""
],
[
"Zhang",
"Y.",
""
],
[
"Zhang",
"C.",
""
]
] |
2504.06521 | Songze Li | Songze Li, Tonghua Su, Xu-Yao Zhang, Qixing Xu, Zhongjie Wang | DUKAE: DUal-level Knowledge Accumulation and Ensemble for Pre-Trained
Model-Based Continual Learning | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Pre-trained model-based continual learning (PTMCL) has garnered growing
attention, as it enables more rapid acquisition of new knowledge by leveraging
the extensive foundational understanding inherent in pre-trained model (PTM).
Most existing PTMCL methods use Parameter-Efficient Fine-Tuning (PEFT) to learn
new knowledge while consolidating existing memory. However, they often face
some challenges. A major challenge lies in the misalignment of classification
heads, as the classification head of each task is trained within a distinct
feature space, leading to inconsistent decision boundaries across tasks and,
consequently, increased forgetting. Another critical limitation stems from the
restricted feature-level knowledge accumulation, with feature learning
typically restricted to the initial task only, which constrains the model's
representation capabilities. To address these issues, we propose a method named
DUal-level Knowledge Accumulation and Ensemble (DUKAE) that leverages both
feature-level and decision-level knowledge accumulation by aligning
classification heads into a unified feature space through Gaussian distribution
sampling and introducing an adaptive expertise ensemble to fuse knowledge
across feature subspaces. Extensive experiments on CIFAR-100, ImageNet-R,
CUB-200, and Cars-196 datasets demonstrate the superior performance of our
approach.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 01:40:38 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 13:22:13 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Li",
"Songze",
""
],
[
"Su",
"Tonghua",
""
],
[
"Zhang",
"Xu-Yao",
""
],
[
"Xu",
"Qixing",
""
],
[
"Wang",
"Zhongjie",
""
]
] |
2504.06571 | Tao Wang | De-hao Zhao, Xiao-shen Kang, Li Gong, Ze-yu Yin, Tao Wang | Double shape quantum phase transitions in the SU3-IBM: new $\gamma$-soft
phase and the shape phase transition from the new $\gamma$-soft phase to the
prolate shape | Comments are welcome | null | null | null | nucl-th nucl-ex | http://creativecommons.org/licenses/by/4.0/ | Shape quantum phase transition is an important topic in nuclear structure. In
this paper, we begin to study the shape quantum phase transition in the
SU3-IBM. In this new proposed model, spherical-like spectra was found to
resolve the spherical nucleus puzzle, which is a new $\gamma$-soft rotational
mode. In this paper, the shape phase transition along the new $\gamma$-soft
line is first discussed, and then the neighbouring case at the prolate side is
also studied. We find that double shape phase transitions occur along a single
parameter path. The new $\gamma$-softness is really a shape phase and the shape
phase transition from the new $\gamma$-soft phase to the prolate shape is
found. The experimental support is also found and $^{108}$Pd may be the
critical nucleus.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 04:15:40 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 08:55:57 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhao",
"De-hao",
""
],
[
"Kang",
"Xiao-shen",
""
],
[
"Gong",
"Li",
""
],
[
"Yin",
"Ze-yu",
""
],
[
"Wang",
"Tao",
""
]
] |
2504.06758 | Tyler Buchanan | Tyler Buchanan, Monica L\u{a}c\u{a}tu\c{s}, Alastair West, and Richard
P. Dwight | Data-Driven RANS Closures Using a Relative Importance Term Analysis
Based Classifier for 2D and 3D Separated Flows | null | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by-nc-sa/4.0/ | This study presents a novel approach for enhancing Reynolds-averaged
Navier-Stokes (RANS) turbulence modeling through the application of a Relative
Importance Term Analysis (RITA) methodology to develop a new zonally-augmented
$k-\omega$ SST model. Traditional Linear Eddy Viscosity Models often struggle
with separated flows. Our approach introduces a physics-based binary classifier
that systematically identifies separated shear layers requiring correction by
analyzing the relative magnitudes of terms in the turbulence kinetic energy
equation. Using symbolic regression, we develop compact correction terms for
Reynolds stress anisotropy and turbulent kinetic energy production. Trained on
two-dimensional configurations, our model demonstrates significant improvements
in predicting separation dynamics while maintaining baseline performance and
fully attached flows. Generalization tests on Ahmed body and Faith Hill
three-dimensional configurations confirm robust transferability, establishing
an effective methodology for targeted enhancement of RANS predictions in
separated flows.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 10:22:36 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 17:07:28 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Buchanan",
"Tyler",
""
],
[
"Lăcătuş",
"Monica",
""
],
[
"West",
"Alastair",
""
],
[
"Dwight",
"Richard P.",
""
]
] |
2504.06818 | Minghao Han | Minghao Han, Xunyuan Yin | Deep Neural Koopman Operator-based Economic Model Predictive Control of
Shipboard Carbon Capture System | null | null | null | null | eess.SY cs.LG cs.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Shipboard carbon capture is a promising solution to help reduce carbon
emissions in international shipping. In this work, we propose a data-driven
dynamic modeling and economic predictive control approach within the Koopman
framework. This integrated modeling and control approach is used to achieve
safe and energy-efficient process operation of shipboard post-combustion carbon
capture plants. Specifically, we propose a deep neural Koopman operator
modeling approach, based on which a Koopman model with time-varying model
parameters is established. This Koopman model predicts the overall economic
operational cost and key system outputs, based on accessible partial state
measurements. By leveraging this learned model, a constrained economic
predictive control scheme is developed. Despite time-varying parameters
involved in the formulated model, the formulated optimization problem
associated with the economic predictive control design is convex, and it can be
solved efficiently during online control implementations. Extensive tests are
conducted on a high-fidelity simulation environment for shipboard
post-combustion carbon capture processes. Four ship operational conditions are
taken into account. The results show that the proposed method significantly
improves the overall economic operational performance and carbon capture rate.
Additionally, the proposed method guarantees safe operation by ensuring that
hard constraints on the system outputs are satisfied.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 12:22:42 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 12:51:12 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Han",
"Minghao",
""
],
[
"Yin",
"Xunyuan",
""
]
] |
2504.06824 | Colin Combe | Juri Rappsilber, James Bruce, Colin Combe, Stephen Fried, Albert J R
Heck, Claudio Iacobucci, Alexander Leitner, Karl Mechtler, Petr Novak,
Francis O'Reilly, David C. Schriemer, Andrea Sinz, Florian Stengel, Andrea
Graziadei, Konstantinos Thalassinos | A Roadmap for Improving Data Reliability and Sharing in Crosslinking
Mass Spectrometry | null | null | null | null | q-bio.OT | http://creativecommons.org/licenses/by/4.0/ | Crosslinking Mass Spectrometry (MS) can uncover protein-protein interactions
and provide structural information on proteins in their native cellular
environments. Despite its promise, the field remains hampered by inconsistent
data formats, variable approaches to error control, and insufficient
interoperability with global data repositories. Recent advances, especially in
false discovery rate (FDR) models and pipeline benchmarking, show that
Crosslinking MS data can reach a reliability that matches the demand of
integrative structural biology. To drive meaningful progress, however, the
community must agree on error estimation, open data formats, and streamlined
repository submissions. This perspective highlights these challenges, clarifies
remaining barriers, and frames practical next steps. Successful field
harmonisation will enhance the acceptance of Crosslinking MS in the broader
biological community and is critical for the dependability of the data, no
matter where it is produced.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 12:32:39 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 12:39:30 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Rappsilber",
"Juri",
""
],
[
"Bruce",
"James",
""
],
[
"Combe",
"Colin",
""
],
[
"Fried",
"Stephen",
""
],
[
"Heck",
"Albert J R",
""
],
[
"Iacobucci",
"Claudio",
""
],
[
"Leitner",
"Alexander",
""
],
[
"Mechtler",
"Karl",
""
],
[
"Novak",
"Petr",
""
],
[
"O'Reilly",
"Francis",
""
],
[
"Schriemer",
"David C.",
""
],
[
"Sinz",
"Andrea",
""
],
[
"Stengel",
"Florian",
""
],
[
"Graziadei",
"Andrea",
""
],
[
"Thalassinos",
"Konstantinos",
""
]
] |
2504.06923 | Emiliano De Cristofaro | Georgi Ganev and Meenatchi Sundaram Muthu Selva Annamalai and Sofiane
Mahiou and Emiliano De Cristofaro | The Importance of Being Discrete: Measuring the Impact of Discretization
in End-to-End Differentially Private Synthetic Data | null | null | null | null | cs.CR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Differentially Private (DP) generative marginal models are often used in the
wild to release synthetic tabular datasets in lieu of sensitive data while
providing formal privacy guarantees. These models approximate low-dimensional
marginals or query workloads; crucially, they require the training data to be
pre-discretized, i.e., continuous values need to first be partitioned into
bins. However, as the range of values (or their domain) is often inferred
directly from the training data, with the number of bins and bin edges
typically defined arbitrarily, this approach can ultimately break end-to-end DP
guarantees and may not always yield optimal utility.
In this paper, we present an extensive measurement study of four
discretization strategies in the context of DP marginal generative models. More
precisely, we design DP versions of three discretizers (uniform, quantile, and
k-means) and reimplement the PrivTree algorithm. We find that optimizing both
the choice of discretizer and bin count can improve utility, on average, by
almost 30% across six DP marginal models, compared to the default strategy and
number of bins, with PrivTree being the best-performing discretizer in the
majority of cases. We demonstrate that, while DP generative models with
non-private discretization remain vulnerable to membership inference attacks,
applying DP during discretization effectively mitigates this risk. Finally, we
propose an optimized approach for automatically selecting the optimal number of
bins, achieving high utility while reducing both privacy budget consumption and
computational overhead.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 14:30:30 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 01:13:47 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ganev",
"Georgi",
""
],
[
"Annamalai",
"Meenatchi Sundaram Muthu Selva",
""
],
[
"Mahiou",
"Sofiane",
""
],
[
"De Cristofaro",
"Emiliano",
""
]
] |
2504.06951 | Christiaan Van De Ven | Christiaan J. F. van de Ven, Muhammad Faisal Khan, S. Serra-Capizzano | GLT hidden structures in mean-field quantum spin systems | 22 pages, 8 figures, 4 tables | null | null | null | quant-ph cs.NA math.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work explores structured matrix sequences arising in mean-field quantum
spin systems. We express these sequences within the framework of generalized
locally Toeplitz (GLT) $*$-algebras, leveraging the fact that each GLT matrix
sequence has a unique GLT symbol. This symbol characterizes both the asymptotic
singular value distribution and, for Hermitian or quasi-Hermitian sequences,
the asymptotic spectral distribution. Specifically, we analyze two cases of
real symmetric matrix sequences stemming from mean-field quantum spin systems
and determine their associated distributions using GLT theory. Our study
concludes with visualizations and numerical tests that validate the theoretical
findings, followed by a discussion of open problems and future directions.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 14:58:31 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 06:44:04 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"van de Ven",
"Christiaan J. F.",
""
],
[
"Khan",
"Muhammad Faisal",
""
],
[
"Serra-Capizzano",
"S.",
""
]
] |
2504.06958 | Xinhao Li | Xinhao Li, Ziang Yan, Desen Meng, Lu Dong, Xiangyu Zeng, Yinan He,
Yali Wang, Yu Qiao, Yi Wang, Limin Wang | VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement
Fine-Tuning | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Recent advancements in reinforcement learning have significantly advanced the
reasoning capabilities of multimodal large language models (MLLMs). While
approaches such as Group Relative Policy Optimization (GRPO) and rule-based
reward mechanisms demonstrate promise in text and image domains, their
application to video understanding remains limited. This paper presents a
systematic exploration of Reinforcement Fine-Tuning (RFT) with GRPO for video
MLLMs, aiming to enhance spatio-temporal perception while maintaining general
capabilities. Our experiments reveal that RFT is highly data-efficient for
task-specific improvements. Through multi-task RFT on spatio-temporal
perception objectives with limited samples, we develop VideoChat-R1, a powerful
video MLLM that achieves state-of-the-art performance on spatio-temporal
perception tasks without sacrificing chat ability, while exhibiting emerging
spatio-temporal reasoning abilities. Compared to Qwen2.5-VL-7B, VideoChat-R1
boosts performance several-fold in tasks like temporal grounding (+31.8) and
object tracking (+31.2). Additionally, it significantly improves on general QA
benchmarks such as VideoMME (+0.9), MVBench (+1.0), and Perception Test (+0.9).
Our findings underscore the potential of RFT for specialized task enhancement
of Video MLLMs. We hope our work offers valuable insights for future RL
research in video MLLMs.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 15:09:27 GMT"
},
{
"version": "v2",
"created": "Thu, 10 Apr 2025 16:28:39 GMT"
},
{
"version": "v3",
"created": "Sun, 13 Apr 2025 09:52:21 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Li",
"Xinhao",
""
],
[
"Yan",
"Ziang",
""
],
[
"Meng",
"Desen",
""
],
[
"Dong",
"Lu",
""
],
[
"Zeng",
"Xiangyu",
""
],
[
"He",
"Yinan",
""
],
[
"Wang",
"Yali",
""
],
[
"Qiao",
"Yu",
""
],
[
"Wang",
"Yi",
""
],
[
"Wang",
"Limin",
""
]
] |
2504.06990 | Francesca Garelli | Francesca Garelli, Erika Michela Dematteis, Vitalie Stavila, Giuseppe
Di Florio, Claudio Carbone, Alessandro Agostini, Mauro Palumbo, Marcello
Baricco, Paola Rizzi | Enhancing TiFe Alloy Activation for Hydrogen Storage Through Al, Cr, Co,
and Cu Substitutions as a Step Towards Future Recycling | null | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by-nc-sa/4.0/ | This study investigates the activation behavior of $TiFe_{0.80}$-$X_{0.20}$
(X = Co, Cu, Cr, Al) alloys to identify the most effective materials for
producing hydrogen storage alloys from recycled sources in view of a circular
economy perspective. Activation was tested using two methods: a Sievert
Volumetric Apparatus at room temperature and 64 bar of hydrogen, and
high-pressure differential scanning calorimetry with 50 bar hydrogen under
thermal cycles up to 400 {\deg}C. Activation properties were analyzed by
assessing time for incubation and for full charging, that are influenced,
respectively, by surface and bulk diffusion of hydrogen. Results showed that
Cr-substituted alloys are rapidly activated, due to the presence of $TiCr_{2}$
compound, while Al-containing alloys absorbed hydrogen immediately. In
contrast, Co- and Cu-substituted alloys required extended activation times, due
to less quantity of secondary phases and limited diffusion channels.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 15:57:51 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 11:16:06 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Garelli",
"Francesca",
""
],
[
"Dematteis",
"Erika Michela",
""
],
[
"Stavila",
"Vitalie",
""
],
[
"Di Florio",
"Giuseppe",
""
],
[
"Carbone",
"Claudio",
""
],
[
"Agostini",
"Alessandro",
""
],
[
"Palumbo",
"Mauro",
""
],
[
"Baricco",
"Marcello",
""
],
[
"Rizzi",
"Paola",
""
]
] |
2504.07005 | Zeyu Liu | Zeyu Liu | A stacky approach to prismatic crystals via $q$-prism charts | null | null | null | null | math.AG math.NT | http://creativecommons.org/licenses/by/4.0/ | Let $Y$ be a locally complete intersection over $\mathcal{O}_K$ containing a
$p$-power root of unity $\zeta_p$. We classify the derived category of
prismatic crystals on the absolute prismatic site of $Y$ by studying
quasi-coherent complexes on the prismatization of $Y$ via $q$-prism charts. We
also develop a Galois descent mechanism to remove the assumption on
$\mathcal{O}_K$. As an application, we classify quasi-coherent complexes on the
Cartier-Witt stack and give a purely algebraic calculation of the cohomology of
the structure sheaf on the absolute prismatic site of $\mathbb{Z}_p$. Along the
way, for $Y$ a locally complete intersection over $\overline{A}$ with $A$ lying
over a $q$-prism, we classify quasi-coherent complexes on the relative
prismatization of $Y$.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 16:24:51 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 05:50:38 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Liu",
"Zeyu",
""
]
] |
2504.07029 | Ran Zhang | Ran Zhang, Xuanhua He, Ke Cao, Liu Liu, Li Zhang, Man Zhou, Jie Zhang | Distilling Textual Priors from LLM to Efficient Image Fusion | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-sa/4.0/ | Multi-modality image fusion aims to synthesize a single, comprehensive image
from multiple source inputs. Traditional approaches, such as CNNs and GANs,
offer efficiency but struggle to handle low-quality or complex inputs. Recent
advances in text-guided methods leverage large model priors to overcome these
limitations, but at the cost of significant computational overhead, both in
memory and inference time. To address this challenge, we propose a novel
framework for distilling large model priors, eliminating the need for text
guidance during inference while dramatically reducing model size. Our framework
utilizes a teacher-student architecture, where the teacher network incorporates
large model priors and transfers this knowledge to a smaller student network
via a tailored distillation process. Additionally, we introduce spatial-channel
cross-fusion module to enhance the model's ability to leverage textual priors
across both spatial and channel dimensions. Our method achieves a favorable
trade-off between computational efficiency and fusion quality. The distilled
network, requiring only 10% of the parameters and inference time of the teacher
network, retains 90% of its performance and outperforms existing SOTA methods.
Extensive experiments demonstrate the effectiveness of our approach. The
implementation will be made publicly available as an open-source resource.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 16:44:19 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 14:47:35 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhang",
"Ran",
""
],
[
"He",
"Xuanhua",
""
],
[
"Cao",
"Ke",
""
],
[
"Liu",
"Liu",
""
],
[
"Zhang",
"Li",
""
],
[
"Zhou",
"Man",
""
],
[
"Zhang",
"Jie",
""
]
] |
2504.07060 | Ruoyu Chen | Ruoyu Chen, Hua Zhang, Jingzhi Li, Li Liu, Zhen Huang, Xiaochun Cao | Generalized Semantic Contrastive Learning via Embedding Side Information
for Few-Shot Object Detection | Accepted by T-PAMI (IEEE Transactions on Pattern Analysis and Machine
Intelligence) | null | 10.1109/TPAMI.2025.3560033 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The objective of few-shot object detection (FSOD) is to detect novel objects
with few training samples. The core challenge of this task is how to construct
a generalized feature space for novel categories with limited data on the basis
of the base category space, which could adapt the learned detection model to
unknown scenarios. However, limited by insufficient samples for novel
categories, two issues still exist: (1) the features of the novel category are
easily implicitly represented by the features of the base category, leading to
inseparable classifier boundaries, (2) novel categories with fewer data are not
enough to fully represent the distribution, where the model fine-tuning is
prone to overfitting. To address these issues, we introduce the side
information to alleviate the negative influences derived from the feature space
and sample viewpoints and formulate a novel generalized feature representation
learning method for FSOD. Specifically, we first utilize embedding side
information to construct a knowledge matrix to quantify the semantic
relationship between the base and novel categories. Then, to strengthen the
discrimination between semantically similar categories, we further develop
contextual semantic supervised contrastive learning which embeds side
information. Furthermore, to prevent overfitting problems caused by sparse
samples, a side-information guided region-aware masked module is introduced to
augment the diversity of samples, which finds and abandons biased information
that discriminates between similar categories via counterfactual explanation,
and refines the discriminative representation space further. Extensive
experiments using ResNet and ViT backbones on PASCAL VOC, MS COCO, LVIS V1,
FSOD-1K, and FSVOD-500 benchmarks demonstrate that our model outperforms the
previous state-of-the-art methods, significantly improving the ability of FSOD
in most shots/splits.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 17:24:05 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Chen",
"Ruoyu",
""
],
[
"Zhang",
"Hua",
""
],
[
"Li",
"Jingzhi",
""
],
[
"Liu",
"Li",
""
],
[
"Huang",
"Zhen",
""
],
[
"Cao",
"Xiaochun",
""
]
] |
2504.07124 | Dukjae Jang | Eunseok Hwang, Myung-Ki Cheoun, Dukjae Jang | Upper Limit of Fusion Reactivity in Laser-Driven $p+{^{11}{\rm B}}$
Reaction | null | null | null | null | physics.plasm-ph | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We explore the averaged fusion reactivity of the $p+{^{11}{\rm B}}$ reaction
in tabletop laser experiments using a plasma expansion model. We investigate
the energy distribution of proton beams accelerated by lasers as a function of
electron temperature $T_e$ and the dimensionless acceleration time $\omega_{pi}
t_{\rm acc}$, where $\omega_{pi}$ is the ion plasma frequency. By combining
these distributions with the fusion cross-section, we identify the optimal
conditions that maximize the fusion reactivity, with $\left\langle \sigma v
\right\rangle = 8.12 \times 10^{-16}\,{\rm cm^3/s}$ at $k_B T_e = 10.0\,{\rm
MeV}$ and $\omega_{pi} t_{\rm acc} = 0.503$. These findings suggest that an
upper limit exists for the fusion reactivity achievable in laser-driven
$p+{^{11}{\rm B}}$ fusion experiments, even under optimized conditions.
| [
{
"version": "v1",
"created": "Mon, 31 Mar 2025 06:24:54 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 06:22:16 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hwang",
"Eunseok",
""
],
[
"Cheoun",
"Myung-Ki",
""
],
[
"Jang",
"Dukjae",
""
]
] |
2504.07207 | Giuseppe Calajo | Giuseppe Calaj\`o, Matija Te\v{c}er, Simone Montangero, Pietro Silvi,
and Marco Di Liberto | Many-body quantum dimerization in 2D atomic arrays | 12 pages 8 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | We consider a 2D atomic array coupled to different photonic environments,
focusing on the half-filled excitation subspace, where strong photon
interactions can give rise to complex many-body states. In particular, we
demonstrate that the least radiant state in this sector is well described by a
coherent superposition of all possible quantum dimer coverings: a resonating
valence bond (RVB) liquid state. We discuss possible strategies to probe this
exotic state, along with their limitations and challenges. Finally, we show
that such a quantum dimer covering can also emerge as the ground state of the
coherent Hamiltonian describing a 2D atomic array coupled to a photonic
band-gap material.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 18:35:24 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 11:48:41 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Calajò",
"Giuseppe",
""
],
[
"Tečer",
"Matija",
""
],
[
"Montangero",
"Simone",
""
],
[
"Silvi",
"Pietro",
""
],
[
"Di Liberto",
"Marco",
""
]
] |
2504.07210 | Paul Borne--Pons | Paul Borne--Pons (Adobe Research, ESA), Mikolaj Czerkawski (Asterisk
Labs, ESA), Rosalie Martin (Adobe Research) and Romain Rouffet (Adobe
Research) | MESA: Text-Driven Terrain Generation Using Latent Diffusion and Global
Copernicus Data | Accepted at CVPR 2025 Workshop MORSE | null | null | null | cs.GR cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Terrain modeling has traditionally relied on procedural techniques, which
often require extensive domain expertise and handcrafted rules. In this paper,
we present MESA - a novel data-centric alternative by training a diffusion
model on global remote sensing data. This approach leverages large-scale
geospatial information to generate high-quality terrain samples from text
descriptions, showcasing a flexible and scalable solution for terrain
generation. The model's capabilities are demonstrated through extensive
experiments, highlighting its ability to generate realistic and diverse terrain
landscapes. The dataset produced to support this work, the Major TOM Core-DEM
extension dataset, is released openly as a comprehensive resource for global
terrain data. The results suggest that data-driven models, trained on remote
sensing data, can provide a powerful tool for realistic terrain modeling and
generation.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 18:37:24 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 17:25:41 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Borne--Pons",
"Paul",
"",
"Adobe Research, ESA"
],
[
"Czerkawski",
"Mikolaj",
"",
"Asterisk\n Labs, ESA"
],
[
"Martin",
"Rosalie",
"",
"Adobe Research"
],
[
"Rouffet",
"Romain",
"",
"Adobe\n Research"
]
] |
2504.07227 | Erik Szwed | E. A. Szwed, B. Vermilyea, D. J. Choksy, Zhiwen Zhou, M. M. Fogler, L.
V. Butov, K. W. Baldwin, L. N. Pfeiffer | Mott transition in excitonic Bose polarons | null | null | null | null | cond-mat.quant-gas | http://creativecommons.org/licenses/by/4.0/ | For a neutral system of positive and negative charges, such as atoms in a
crystal, increasing the density causes the Mott transition from bound electrons
to free electrons. The density of optically generated electron-hole systems can
be controlled in situ by the power of optical excitation that enables the Mott
transition from excitons, the bound pairs of electrons and holes, to free
electrons and holes with increasing density. These Mott transitions occur in
systems of pairs of the same kind, such as atoms or excitons. However, a
different type of the Mott transition can occur for Bose polarons. A Bose
polaron is a mobile particle of one kind in a Bose gas of particles of another
kind. For the Mott transition in polarons, the polaron states vanish with
increasing density of the surrounding gas. In this paper, we present the
observation of this type of the Mott transition and the measurement of the Mott
transition parameter $n_{\rm M}^{1/2} a_{\rm B}$ in 2D excitonic Bose polarons.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 19:08:28 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 19:31:43 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Szwed",
"E. A.",
""
],
[
"Vermilyea",
"B.",
""
],
[
"Choksy",
"D. J.",
""
],
[
"Zhou",
"Zhiwen",
""
],
[
"Fogler",
"M. M.",
""
],
[
"Butov",
"L. V.",
""
],
[
"Baldwin",
"K. W.",
""
],
[
"Pfeiffer",
"L. N.",
""
]
] |
2504.07233 | Adil Bahaj | Yousra Fettach and Adil Bahaj and Mounir Ghogho | Skill Demand Forecasting Using Temporal Knowledge Graph Embeddings | 17 pages | null | null | null | cs.CY | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Rapid technological advancements pose a significant threat to a large portion
of the global workforce, potentially leaving them behind. In today's economy,
there is a stark contrast between the high demand for skilled labour and the
limited employment opportunities available to those who are not adequately
prepared for the digital economy. To address this critical juncture and gain a
deeper and more rapid understanding of labour market dynamics, in this paper,
we approach the problem of skill need forecasting as a knowledge graph (KG)
completion task, specifically, temporal link prediction. We introduce our novel
temporal KG constructed from online job advertisements. We then train and
evaluate different temporal KG embeddings for temporal link prediction.
Finally, we present predictions of demand for a selection of skills practiced
by workers in the information technology industry. The code and the data are
available on our GitHub repository https://github.com/team611/JobEd.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 19:18:29 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 10:43:04 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Fettach",
"Yousra",
""
],
[
"Bahaj",
"Adil",
""
],
[
"Ghogho",
"Mounir",
""
]
] |
2504.07238 | Christopher Sogge | Xiaoqi Huang, Christopher D. Sogge, Zhongkai Tao and Zhexing Zhang | Lossless Strichartz and spectral projection estimates on unbounded
manifolds | 69 pages, corrected title | null | null | null | math.AP math.CA math.SP | http://creativecommons.org/licenses/by/4.0/ | We prove new lossless Strichartz and spectral projection estimates on
asymptotically hyperbolic surfaces, and, in particular, on all convex cocompact
hyperbolic surfaces. In order to do this, we also obtain log-scale lossless
Strichartz and spectral projection estimates on manifolds of uniformly bounded
geometry with nonpositive and negative sectional curvatures, extending the
recent works of the first two authors for compact manifolds. We are able to use
these along with known $L^2$-local smoothing and new $L^2 \to L^q$
half-localized resolvent estimates to obtain our lossless bounds.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 19:26:03 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 15:00:30 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Huang",
"Xiaoqi",
""
],
[
"Sogge",
"Christopher D.",
""
],
[
"Tao",
"Zhongkai",
""
],
[
"Zhang",
"Zhexing",
""
]
] |
2504.07249 | Natan Spear | Kim H. Pham, Vijaya Begum-Hudde, Amy K. Lin, Natan A. Spear, Jackson
McClellan, Michael W. Zuerch, Andre Schleife, Kimberly A. See, and Scott
Cushing | The dynamical role of optical phonons and sub-lattice screening in a
solid-state ion conductor | 7 figures, 1 table | null | null | null | cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Solid-state electrolytes (SSEs) require ionic conductivities that are
competitive with liquid electrolytes to realize applications in all-solid state
batteries. Although numerous materials have been discovered, the underlying
mechanisms enabling superionic conduction remain elusive. In particular, the
role of ultrafast lattice dynamics in mediating ion migration, which involves
couplings between ions, phonons, and electrons, is rarely explored
experimentally at their corresponding timescales. To investigate the
contributions of coupled lattice dynamics on ion migration, we modulate the
charge density occupations within the crystal, and then measure the
time-resolved change in impedance on picosecond timescales for a candidate SSE,
Li0.5La0.5TiO3. Upon perturbation, we observe enhanced ion migration at
ultrafast timescales that are shorter than laser-induced heating. The
respective transients match the timescales of optical and acoustic phonon
vibrations, suggesting their involvement in ion migration. We further
computationally evaluate the effect of a charge transfer from the O 2p to Ti 3d
band on the electronic and physical structure of LLTO. We hypothesize that the
charge transfer excitation distorts the TiO6 polyhedra by altering the local
charge density occupancy of the hopping site at the migration pathway saddle
point, thereby causing a reduction in the migration barrier for the Li+ hop,
shown using computation. We rule out the contribution of photogenerated
electrons and laser heating. Overall, our investigation introduces a new
spectroscopic tool to probe fundamental ion hopping mechanisms transiently at
ultrafast timescales, which has previously only been achieved in a
time-averaged manner or solely via computational methods. Our proposed
technique expands our capability to answer dynamical questions previously
limited by incumbent spectroscopic strategies.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 19:42:57 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 19:11:08 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Pham",
"Kim H.",
""
],
[
"Begum-Hudde",
"Vijaya",
""
],
[
"Lin",
"Amy K.",
""
],
[
"Spear",
"Natan A.",
""
],
[
"McClellan",
"Jackson",
""
],
[
"Zuerch",
"Michael W.",
""
],
[
"Schleife",
"Andre",
""
],
[
"See",
"Kimberly A.",
""
],
[
"Cushing",
"Scott",
""
]
] |
2504.07253 | Erin Sheridan | Erin Sheridan, Michael Senatore, Samuel Schwab, Eric Aspling, Taylor
Wagner, James Schneeloch, Stephen McCoy, Daniel Campbell, David Hucul,
Zachary Smith, Matthew LaHaye | Noise-Aware Entanglement Generation Protocols for Superconducting Qubits
with Impedance-Matched FBAR Transducers | Fixed typos and updated Figures 6-10 | null | null | null | quant-ph cond-mat.mes-hall | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Connecting superconducting quantum processors to
telecommunications-wavelength quantum networks is critically necessary to
enable distributed quantum computing, secure communications, and other
applications. Optically-mediated entanglement heralding protocols offer a
near-term solution that can succeed with imperfect components, including
sub-unity efficiency microwave-optical quantum transducers. The viability and
performance of these protocols relies heavily on the properties of the
transducers used: the conversion efficiency, resonator lifetimes, and added
noise in the transducer directly influence the achievable entanglement
generation rate and fidelity of an entanglement generation protocol. Here, we
use an extended Butterworth-van Dyke (BVD) model to optimize the conversion
efficiency and added noise of a Thin Film Bulk Acoustic Resonator (FBAR)
piezo-optomechanical transducer. We use the outputs from this model to
calculate the fidelity of one-photon and two-photon entanglement heralding
protocols in a variety of operating regimes. For transducers with matching
circuits designed to either minimize the added noise or maximize conversion
efficiency, we theoretically estimate that entanglement generation rates of
greater than $160\;\mathrm{kHz}$ can be achieved at moderate pump powers with
fidelities of $>90\%$. This is the first time a BVD equivalent circuit model is
used to both optimize the performance of an FBAR transducer and to directly
inform the design and implementation of an entanglement generation protocol.
These results can be applied in the near term to realize quantum networks of
superconducting qubits with realistic experimental parameters.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 20:10:04 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 23:35:01 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Sheridan",
"Erin",
""
],
[
"Senatore",
"Michael",
""
],
[
"Schwab",
"Samuel",
""
],
[
"Aspling",
"Eric",
""
],
[
"Wagner",
"Taylor",
""
],
[
"Schneeloch",
"James",
""
],
[
"McCoy",
"Stephen",
""
],
[
"Campbell",
"Daniel",
""
],
[
"Hucul",
"David",
""
],
[
"Smith",
"Zachary",
""
],
[
"LaHaye",
"Matthew",
""
]
] |
2504.07282 | Yangning Li | Lv Qingsong, Yangning Li, Zihua Lan, Zishan Xu, Jiwei Tang, Yinghui
Li, Wenhao Jiang, Hai-Tao Zheng, Philip S. Yu | RAISE: Reinforenced Adaptive Instruction Selection For Large Language
Models | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the instruction fine-tuning of large language models (LLMs), it has become
a consensus that a few high-quality instructions are superior to a large number
of low-quality instructions. At present, many instruction selection methods
have been proposed, but most of these methods select instruction based on
heuristic quality metrics, and only consider data selection before training.
These designs lead to insufficient optimization of instruction fine-tuning, and
fixed heuristic indicators are often difficult to optimize for specific tasks.
So we designed a dynamic, task-objective-driven instruction selection framework
RAISE(Reinforenced Adaptive Instruction SElection), which incorporates the
entire instruction fine-tuning process into optimization, selecting instruction
at each step based on the expected impact of instruction on model performance
improvement. Our approach is well interpretable and has strong task-specific
optimization capabilities. By modeling dynamic instruction selection as a
sequential decision-making process, we use RL to train our selection strategy.
Extensive experiments and result analysis prove the superiority of our method
compared with other instruction selection methods. Notably, RAISE achieves
superior performance by updating only 1\% of the training steps compared to
full-data training, demonstrating its efficiency and effectiveness.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 21:17:52 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 16:23:29 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Qingsong",
"Lv",
""
],
[
"Li",
"Yangning",
""
],
[
"Lan",
"Zihua",
""
],
[
"Xu",
"Zishan",
""
],
[
"Tang",
"Jiwei",
""
],
[
"Li",
"Yinghui",
""
],
[
"Jiang",
"Wenhao",
""
],
[
"Zheng",
"Hai-Tao",
""
],
[
"Yu",
"Philip S.",
""
]
] |
2504.07288 | Yangning Li | Yangning Li, Zihua Lan, Lv Qingsong, Yinghui Li, Hai-Tao Zheng | MDIT: A Model-free Data Interpolation Method for Diverse Instruction
Tuning | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As Large Language Models (LLMs) are increasingly applied across various
tasks, instruction tuning has emerged as a critical method for enhancing model
performance. However, current data management strategies face substantial
challenges in generating diverse and comprehensive data, restricting further
improvements in model performance. To address this gap, we propose MDIT, a
novel model-free data interpolation method for diverse instruction tuning,
which generates varied and high-quality instruction data by performing task
interpolation. Moreover, it contains diversity-based clustering strategies to
ensure the diversity of the training data. Extensive experiments show that our
method achieves superior performance in multiple benchmark tasks. The LLMs
finetuned with MDIT show significant improvements in numerous tasks such as
general question answering, math reasoning, and code generation. MDIT offers an
efficient and automatic data synthetic method, generating diverse instruction
data without depending on external resources while expanding the application
potential of LLMs in complex environments.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 21:28:17 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 17:48:08 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Li",
"Yangning",
""
],
[
"Lan",
"Zihua",
""
],
[
"Qingsong",
"Lv",
""
],
[
"Li",
"Yinghui",
""
],
[
"Zheng",
"Hai-Tao",
""
]
] |
2504.07312 | Genevieve Smith | Genevieve Smith | The Gendered Algorithm: Navigating Financial Inclusion & Equity in
AI-facilitated Access to Credit | null | null | null | null | cs.CY | http://creativecommons.org/licenses/by/4.0/ | A growing trend in financial technology (fintech) is the use of mobile phone
data and machine learning (ML) to provide credit scores- and subsequently,
opportunities to access loans- to groups left out of traditional banking. This
paper draws on interview data with leaders, investors, and data scientists at
fintech companies developing ML-based alternative lending apps in low- and
middle-income countries to explore financial inclusion and gender implications.
More specifically, it examines how the underlying logics, design choices, and
management decisions of ML-based alternative lending tools by fintechs embed or
challenge gender biases, and consequently influence gender equity in access to
finance. Findings reveal developers follow 'gender blind' approaches, grounded
in beliefs that ML is objective and data reflects the truth. This leads to a
lack of grappling with the ways data, features for creditworthiness, and access
to apps are gendered. Overall, tools increase access to finance, but not gender
equitably: Interviewees report less women access loans and receive lower
amounts than men, despite being better repayers. Fintechs identify demand- and
supply-side reasons for gender differences, but frame them as outside their
responsibility. However, that women are observed as better repayers reveals a
market inefficiency and potential discriminatory effect, further linked to
profit optimization objectives. This research introduces the concept of encoded
gender norms, whereby without explicit attention to the gendered nature of data
and algorithmic design, AI tools reproduce existing inequalities. In doing so,
they reinforce gender norms as self-fulfilling prophecies. The idea that AI is
inherently objective and, when left alone, 'fair', is seductive and misleading.
In reality, algorithms reflect the perspectives, priorities, and values of the
people and institutions that design them.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 22:28:21 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 21:16:09 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Smith",
"Genevieve",
""
]
] |
2504.07316 | Shujin Wu | Shujin Wu, Cheng Qian, Yi R. Fung, Paul Pu Liang, Heng Ji | Alice: Proactive Learning with Teacher's Demonstrations for
Weak-to-Strong Generalization | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The growing capabilities of large language models (LLMs) present a key
challenge of maintaining effective human oversight. Weak-to-strong
generalization (W2SG) offers a promising framework for supervising increasingly
capable LLMs using weaker ones. Traditional W2SG methods rely on passive
learning, where a weak teacher provides noisy demonstrations to train a strong
student. This hinders students from employing their knowledge during training
and reaching their full potential. In this work, we introduce Alice
(pro{A}ctive {l}earning w{i}th tea{c}her's D{e}monstrations), a framework that
leverages complementary knowledge between teacher and student to enhance the
learning process. We probe the knowledge base of the teacher model by eliciting
their uncertainty, and then use these insights together with teachers'
responses as demonstrations to guide student models in self-generating improved
responses for supervision. In addition, for situations with significant
capability gaps between teacher and student models, we introduce cascade Alice,
which employs a hierarchical training approach where weak teachers initially
supervise intermediate models, who then guide stronger models in sequence.
Experimental results demonstrate that our method significantly enhances the
W2SG performance, yielding substantial improvements in three key tasks compared
to the original W2SG: knowledge-based reasoning (+4.0%), mathematical reasoning
(+22.62%), and logical reasoning (+12.11%). This highlights the effectiveness
of our new W2SG paradigm that enables more robust knowledge transfer and
supervision outcome.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 22:33:06 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 20:13:23 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Wu",
"Shujin",
""
],
[
"Qian",
"Cheng",
""
],
[
"Fung",
"Yi R.",
""
],
[
"Liang",
"Paul Pu",
""
],
[
"Ji",
"Heng",
""
]
] |
2504.07334 | Cindy Le | Chendi Lin, Heshan Liu, Qunshu Lin, Zachary Bright, Shitao Tang, Yihui
He, Minghao Liu, Ling Zhu, Cindy Le | Objaverse++: Curated 3D Object Dataset with Quality Annotations | 8 pages, 8 figures. Accepted to CVPR 2025 Workshop on Efficient Large
Vision Models (April 2025) | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | This paper presents Objaverse++, a curated subset of Objaverse enhanced with
detailed attribute annotations by human experts. Recent advances in 3D content
generation have been driven by large-scale datasets such as Objaverse, which
contains over 800,000 3D objects collected from the Internet. Although
Objaverse represents the largest available 3D asset collection, its utility is
limited by the predominance of low-quality models. To address this limitation,
we manually annotate 10,000 3D objects with detailed attributes, including
aesthetic quality scores, texture color classifications, multi-object
composition flags, transparency characteristics, etc. Then, we trained a neural
network capable of annotating the tags for the rest of the Objaverse dataset.
Through experiments and a user study on generation results, we demonstrate that
models pre-trained on our quality-focused subset achieve better performance
than those trained on the larger dataset of Objaverse in image-to-3D generation
tasks. In addition, by comparing multiple subsets of training data filtered by
our tags, our results show that the higher the data quality, the faster the
training loss converges. These findings suggest that careful curation and rich
annotation can compensate for the raw dataset size, potentially offering a more
efficient path to develop 3D generative models. We release our enhanced dataset
of approximately 500,000 curated 3D models to facilitate further research on
various downstream tasks in 3D computer vision. In the near future, we aim to
extend our annotations to cover the entire Objaverse dataset.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 23:29:08 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 23:48:12 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Lin",
"Chendi",
""
],
[
"Liu",
"Heshan",
""
],
[
"Lin",
"Qunshu",
""
],
[
"Bright",
"Zachary",
""
],
[
"Tang",
"Shitao",
""
],
[
"He",
"Yihui",
""
],
[
"Liu",
"Minghao",
""
],
[
"Zhu",
"Ling",
""
],
[
"Le",
"Cindy",
""
]
] |
2504.07386 | Fangfang Du | Fang-Fang Du, Ling-Hui Li, Qiu-Lin Tan, Zhuo-Ya Bai | Heralded qudit-based high-dimensional entanglement generation for hybrid
photon-emitter system by waveguide-mediated scattering | null | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Quantum entanglement systems based on qudits dilate high-dimensional (HD)
state spaces and enhance resistance to loss in quantum information processing
(QIP). To fully exploit this potential, effective schemes for generating HD
entanglement are crucial. In this paper, we propose a flexible heralded scheme
generating random 4D two-qudit maximal entanglement for hybrid photon-emitter
system by entering different input ports. This approach can be further extended
to prepare 4D n-qudit (n is greater than or equal 3) maximal entanglement
utilizing the 4D single-qudit Z^m (m=1,2,3) gate for the first qudit and X^m
gate for the other qudits (except the second qudit). For the hybrid system, the
first 4D qudit is encoded on the hybrid polarization-path states of a flying
photon, while the second and subsequent 4D qudits are represented by two
stationary emitters coupled to respective 1D waveguide. The qudit-encoded
hybrid HD entanglement offers advantages over economizing quantum resource
without any auxiliary qudits, and obtaining robust fidelities of various HD
entanglement by the error-detected mechanism of the emitter-waveguide systems.
Moreover, the proposed protocol can be spread to generate dD n-qudit (d is
greater than or equal 2^(p+1), n, p=2,3,...) entangled states, further
broadening its applicability in HD QIP.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 02:12:41 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 04:30:06 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Du",
"Fang-Fang",
""
],
[
"Li",
"Ling-Hui",
""
],
[
"Tan",
"Qiu-Lin",
""
],
[
"Bai",
"Zhuo-Ya",
""
]
] |
2504.07431 | Shavbo Salehi | Shavbo Salehi, Melike Erol-Kantarci, Dusit Niyato | LLM-Enabled Data Transmission in End-to-End Semantic Communication | null | null | null | null | cs.NI | http://creativecommons.org/licenses/by/4.0/ | Emerging services such as augmented reality (AR) and virtual reality (VR)
have increased the volume of data transmitted in wireless communication
systems, revealing the limitations of traditional Shannon theory. To address
these limitations, semantic communication has been proposed as a solution that
prioritizes the meaning of messages over the exact transmission of bits. This
paper explores semantic communication for text data transmission in end-to-end
(E2E) systems through a novel approach called KG-LLM semantic communication,
which integrates knowledge graph (KG) extraction and large language model (LLM)
coding. In this method, the transmitter first utilizes a KG to extract key
entities and relationships from sentences. The extracted information is then
encoded using an LLM to obtain the semantic meaning. On the receiver side,
messages are decoded using another LLM, while a bidirectional encoder
representations from transformers (i.e., BERT) model further refines the
reconstructed sentences for improved semantic similarity. The KG-LLM semantic
communication method reduces the transmitted text data volume by 30% through
KG-based compression and achieves 84\% semantic similarity between the original
and received messages. This demonstrates the KG-LLM methods efficiency and
robustness in semantic communication systems, outperforming the deep
learning-based semantic communication model (DeepSC), which achieves only 63%.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 04:00:11 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 03:40:56 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Salehi",
"Shavbo",
""
],
[
"Erol-Kantarci",
"Melike",
""
],
[
"Niyato",
"Dusit",
""
]
] |
2504.07510 | Shujie Cheng | Feng Lu, Hailin Tan, Ao Zhou, Shujie Cheng, Gao Xianlong | Wigner distribution, Wigner entropy, and Anomalous Transport of a
Generalized Aubry-Andr\'{e} model | 7 pages, 5 figures | null | null | null | cond-mat.dis-nn | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we study a generalized Aubry-Andr\'{e} model with tunable
quasidisordered potentials. The model has an invariable mobility edge that
separates the extended states from the localized states. At the mobility edge,
the wave function presents critical characteristics, which can be verified by
finite-size scaling analysis. Our numerical investigations demonstrate that the
extended, critical, and localized states can be effectively distinguished via
their phase space representation, specially the Wigner distribution. Based on
the Wigner distribution function, we can further obtain the corresponding
Wigner entropy and employ the feature that the critical state has the maximum
Wigner entropy to locate the invariable mobility edge. Finally, we reveal that
there are anomalous transport phenomena between the transition from ballistic
transport to the absence of diffusion.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 07:12:17 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 17:12:55 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Lu",
"Feng",
""
],
[
"Tan",
"Hailin",
""
],
[
"Zhou",
"Ao",
""
],
[
"Cheng",
"Shujie",
""
],
[
"Xianlong",
"Gao",
""
]
] |
2504.07523 | Yizhi Luo | Yizhi Luo, Hilel Hagai Diamandi, Hanshi Li, Runjiang Bi, David Mason,
Taekwan Yoon, Xinghan Guo, Hanlin Tang, Ryan O. Behunin, Frederick J. Walker,
Charles Ahn, Peter T. Rakich | Lifetime-limited Gigahertz-frequency Mechanical Oscillators with
Millisecond Coherence Times | 23 pages, 10 figures | null | null | null | quant-ph cond-mat.mtrl-sci | http://creativecommons.org/licenses/by-nc-nd/4.0/ | High-frequency mechanical oscillators with long coherence times are essential
to realizing a variety of high-fidelity quantum sensors, transducers, and
memories. However, the unprecedented coherence times needed for quantum
applications require exquisitely sensitive new techniques to probe the material
origins of phonon decoherence and new strategies to mitigate decoherence in
mechanical oscillators. Here, we combine non-invasive laser spectroscopy
techniques with materials analysis to identify key sources of phonon
decoherence in crystalline media. Using micro-fabricated high-overtone bulk
acoustic-wave resonators ($\mu$HBARs) as an experimental testbed, we identify
phonon-surface interactions as the dominant source of phonon decoherence in
crystalline quartz; lattice distortion, subsurface damage, and high
concentration of elemental impurities near the crystal surface are identified
as the likely causes. Removal of this compromised surface layer using an
optimized polishing process is seen to greatly enhance coherence times,
enabling $\mu$HBARs with Q-factors of > 240 million at 12 GHz frequencies,
corresponding to > 6 ms phonon coherence times and record-level f-Q products.
Complementary phonon linewidth and time-domain ringdown measurements, performed
using a new Brillouin-based pump-probe spectroscopy technique, reveal
negligible dephasing within these oscillators. Building on these results, we
identify a path to > 100 ms coherence times as the basis for high-frequency
quantum memories. These findings clearly demonstrate that, with enhanced
control over surfaces, dissipation and noise can be significantly reduced in a
wide range of quantum systems.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 07:41:47 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 03:36:37 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Luo",
"Yizhi",
""
],
[
"Diamandi",
"Hilel Hagai",
""
],
[
"Li",
"Hanshi",
""
],
[
"Bi",
"Runjiang",
""
],
[
"Mason",
"David",
""
],
[
"Yoon",
"Taekwan",
""
],
[
"Guo",
"Xinghan",
""
],
[
"Tang",
"Hanlin",
""
],
[
"Behunin",
"Ryan O.",
""
],
[
"Walker",
"Frederick J.",
""
],
[
"Ahn",
"Charles",
""
],
[
"Rakich",
"Peter T.",
""
]
] |
2504.07535 | Yuji Muta | Tatsuya Kataoka, Yuji Muta, Naoki Terai | The v-numbers of Stanley-Reisner ideals from the viewpoint of Alexander
dual complexes | 20 pages | null | null | null | math.AC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We express the v-number of the Stanley-Reisner ideal in terms of its
Alexander dual complex and prove that the v-number of a cover ideal is just two
less than the initial degree of the its syzygy module. We give some relation
between the v-number of the Stanley-Reisner ideal and the Serre-depth of the
quotient ring of the second symbolic power of the Stanley-Reisner ideal of its
Alexander dual. We also show that the v-number of the Stanley-Reisner ideal of
a 2-pure simplicial complex is equal to the dimension of its Stanley-Reisner
ring.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 08:01:59 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 13:11:22 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Kataoka",
"Tatsuya",
""
],
[
"Muta",
"Yuji",
""
],
[
"Terai",
"Naoki",
""
]
] |
2504.07543 | Jinwoo Kim Prof. | Minjae Seo, Myoungsung You, Jaehan Kim, Taejune Park, Seungwon Shin,
Jinwoo Kim | MUFFLER: Secure Tor Traffic Obfuscation with Dynamic Connection
Shuffling and Splitting | To appear in IEEE INFOCOM 2025 | null | null | null | cs.CR cs.NI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Tor, a widely utilized privacy network, enables anonymous communication but
is vulnerable to flow correlation attacks that deanonymize users by correlating
traffic patterns from Tor's ingress and egress segments. Various defenses have
been developed to mitigate these attacks; however, they have two critical
limitations: (i) significant network overhead during obfuscation and (ii) a
lack of dynamic obfuscation for egress segments, exposing traffic patterns to
adversaries. In response, we introduce MUFFLER, a novel connection-level
traffic obfuscation system designed to secure Tor egress traffic. It
dynamically maps real connections to a distinct set of virtual connections
between the final Tor nodes and targeted services, either public or hidden.
This approach creates egress traffic patterns fundamentally different from
those at ingress segments without adding intentional padding bytes or timing
delays. The mapping of real and virtual connections is adjusted in real-time
based on ongoing network conditions, thwarting adversaries' efforts to detect
egress traffic patterns. Extensive evaluations show that MUFFLER mitigates
powerful correlation attacks with a TPR of 1% at an FPR of 10^-2 while imposing
only a 2.17% bandwidth overhead. Moreover, it achieves up to 27x lower latency
overhead than existing solutions and seamlessly integrates with the current Tor
architecture.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 08:17:17 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 13:11:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Seo",
"Minjae",
""
],
[
"You",
"Myoungsung",
""
],
[
"Kim",
"Jaehan",
""
],
[
"Park",
"Taejune",
""
],
[
"Shin",
"Seungwon",
""
],
[
"Kim",
"Jinwoo",
""
]
] |
2504.07568 | Badie Ghavami | Badie Ghavami, Forouzan Mirmasoudi | Ground State Energy of He molecule Using a Four-Qubit Photonic Processor
with the Variational Quantum Eigensolver | 7 pages, 3 figures, 1 table | null | null | null | quant-ph cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | To understand the properties and interactions of materials, and determining
the ground state energies is one of the important challenges in quantum
chemistry, materials science, and quantum mechanics, where quantum computing
can play an important role for studying the properties of materials. In this
study, we have explored the quantum processor application to compute the He
molecule ground state energy which utilizes the Variational Quantum Eigensolver
(VQE) algorithm. In here, we have implemented VQE on a state-of-the-art quantum
processor, optimizing a parameterized quantum circuit to minimize the energy
expectation value of the He molecule's Hamiltonian on the four qubits
processor. The obtained results of this work show a significant improvement in
accuracy compared to classical computational methods, such as Hartree-Fock and
density functional theory, which demonstrate the compute potential of quantum
algorithms in quantum many-body problems. Thus, these results demonstrate the
advantages of quantum computing in achieving high accuracy in simulations of
molecular and material properties, and pave the way for future applications in
more complex systems. This work highlights the potential of quantum processors
in the fields of quantum chemistry, computational physics, and data science.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 09:00:08 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 12:23:11 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ghavami",
"Badie",
""
],
[
"Mirmasoudi",
"Forouzan",
""
]
] |
2504.07615 | Tiancheng Zhao | Haozhan Shen, Peng Liu, Jingcheng Li, Chunxin Fang, Yibo Ma, Jiajia
Liao, Qiaoli Shen, Zilun Zhang, Kangjia Zhao, Qianqian Zhang, Ruochen Xu,
Tiancheng Zhao | VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model | 11 pages, fix some minor typos in the previous version | null | null | null | cs.CV cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Recently DeepSeek R1 has shown that reinforcement learning (RL) can
substantially improve the reasoning capabilities of Large Language Models
(LLMs) through a simple yet effective design. The core of R1 lies in its
rule-based reward formulation, which leverages tasks with deterministic
ground-truth answers to enable precise and stable reward computation. In the
visual domain, we similarly observe that a wide range of visual understanding
tasks are inherently equipped with well-defined ground-truth annotations. This
property makes them naturally compatible with rule-based reward mechanisms.
Motivated by this observation, we investigate the extension of R1-style
reinforcement learning to Vision-Language Models (VLMs), aiming to enhance
their visual reasoning capabilities. To this end, we develop VLM-R1, a
dedicated framework designed to harness RL for improving VLMs' performance on
general vision-language tasks. Using this framework, we further explore the
feasibility of applying RL to visual domain. Experimental results indicate that
the RL-based model not only delivers competitive performance on visual
understanding tasks but also surpasses Supervised Fine-Tuning (SFT) in
generalization ability. Furthermore, we conduct comprehensive ablation studies
that uncover a series of noteworthy insights, including the presence of reward
hacking in object detection, the emergence of the "OD aha moment", the impact
of training data quality, and the scaling behavior of RL across different model
sizes. Through these analyses, we aim to deepen the understanding of how
reinforcement learning enhances the capabilities of vision-language models, and
we hope our findings and open-source contributions will support continued
progress in the vision-language RL community. Our code and model are available
at https://github.com/om-ai-lab/VLM-R1
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 10:05:15 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 15:15:54 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Shen",
"Haozhan",
""
],
[
"Liu",
"Peng",
""
],
[
"Li",
"Jingcheng",
""
],
[
"Fang",
"Chunxin",
""
],
[
"Ma",
"Yibo",
""
],
[
"Liao",
"Jiajia",
""
],
[
"Shen",
"Qiaoli",
""
],
[
"Zhang",
"Zilun",
""
],
[
"Zhao",
"Kangjia",
""
],
[
"Zhang",
"Qianqian",
""
],
[
"Xu",
"Ruochen",
""
],
[
"Zhao",
"Tiancheng",
""
]
] |
2504.07633 | Akira Tamamori | Akira Tamamori | Kernel Logistic Regression Learning for High-Capacity Hopfield Networks | submitted to IEICE journal | null | null | null | cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Hebbian learning limits Hopfield network storage capacity (pattern-to-neuron
ratio around 0.14). We propose Kernel Logistic Regression (KLR) learning.
Unlike linear methods, KLR uses kernels to implicitly map patterns to
high-dimensional feature space, enhancing separability. By learning dual
variables, KLR dramatically improves storage capacity, achieving perfect recall
even when pattern numbers exceed neuron numbers (up to ratio 1.5 shown), and
enhances noise robustness. KLR demonstrably outperforms Hebbian and linear
logistic regression approaches.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 10:27:43 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 00:29:35 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Tamamori",
"Akira",
""
]
] |
2504.07709 | Zheng Zhang | Zheng Zhang, Yuanwei Liu, Bingtao He, Jian Chen | Integrated Sensing and Communications for Pinching-Antenna Systems
(PASS) | 5 pages | null | null | null | cs.IT math.IT | http://creativecommons.org/licenses/by/4.0/ | An integrated sensing and communication (ISAC) design for pinching antenna
systems (PASS) is proposed, where the pinching antennas are deployed for
establishing reliable line-of-sight communication and sensing links. More
particularly, a separated ISAC design is proposed for the two-waveguide PASS,
where one waveguide is used to emit the joint communication and sensing signals
while the other waveguide is used to receive the reflected echo signals. Based
on this framework, a penalty-based alternating optimization algorithm is
proposed to maximize the illumination power as well as ensure the communication
quality-of-service requirement. Numerical results demonstrate that 1) the
proposed PASS-ISAC scheme outperforms the other baseline schemes, and 2) the
considered equal power allocation model achieves an upper bound performance.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 12:58:46 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 17:29:06 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhang",
"Zheng",
""
],
[
"Liu",
"Yuanwei",
""
],
[
"He",
"Bingtao",
""
],
[
"Chen",
"Jian",
""
]
] |
2504.07720 | Juan Manuel Miramont | Juan Manuel Miramont, Kin Aun Tan, Soumendu Sundar Mukherjee, R\'emi
Bardenet, Subhroshekhar Ghosh | Filtering through a topological lens: homology for point processes on
the time-frequency plane | null | null | null | null | eess.SP math.AT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a very general approach to the analysis of signals from their
noisy measurements from the perspective of Topological Data Analysis (TDA).
While TDA has emerged as a powerful analytical tool for data with pronounced
topological structures, here we demonstrate its applicability for general
problems of signal processing, without any a-priori geometric feature. Our
methods are well-suited to a wide array of time-dependent signals in different
scientific domains, with acoustic signals being a particularly important
application. We invoke time-frequency representations of such signals, focusing
on their zeros which are gaining salience as a signal processing tool in view
of their stability properties. Leveraging state-of-the-art topological
concepts, such as stable and minimal volumes, we develop a complete suite of
TDA-based methods to explore the delicate stochastic geometry of these zeros,
capturing signals based on the disruption they cause to this rigid,
hyperuniform spatial structure. Unlike classical spatial data tools, TDA is
able to capture the full spectrum of the stochastic geometry of the zeros,
thereby leading to powerful inferential outcomes that are underpinned by a
principled statistical foundation. This is reflected in the power and
versatility of our applications, which include competitive performance in
processing. a wide variety of audio signals (esp. in low SNR regimes),
effective detection and reconstruction of gravitational wave signals (a reputed
signal processing challenge with non-Gaussian noise), and medical time series
data from EEGs, indicating a wide horizon for the approach and methods
introduced in this paper.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 13:10:04 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 02:45:54 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Miramont",
"Juan Manuel",
""
],
[
"Tan",
"Kin Aun",
""
],
[
"Mukherjee",
"Soumendu Sundar",
""
],
[
"Bardenet",
"Rémi",
""
],
[
"Ghosh",
"Subhroshekhar",
""
]
] |
2504.07746 | ChiYi Luo | Chiyi Luo and Dawei Yang | Upper semi-continuity of metric entropy for $\mathcal{C}^{1,\alpha}$
diffeomorphisms | 34pages, Added an important comment on page 3 | null | null | null | math.DS | http://creativecommons.org/publicdomain/zero/1.0/ | We prove that for $\mathcal{C}^{1,\alpha}$ diffeomorphisms on a compact
manifold $M$ with ${\rm dim} M\leq 3$, if an invariant measure $\mu$ is a
continuity point of the sum of positive Lyapunov exponents, then $\mu$ is an
upper semi-continuity point of the entropy map. This gives several
consequences, such as the upper-semi continuity of dimensions of measures for
surface diffeomorphisms. Furthermore, we know the continuity of dimensions for
measures of maximal entropy.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 13:41:37 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 01:01:37 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Luo",
"Chiyi",
""
],
[
"Yang",
"Dawei",
""
]
] |
2504.07789 | Francisco Guzman | F. Guzm\'an, M. Chatzikos and G. Ferland | H-, He-like recombination spectra -- V: On the dependence of the
simulated line intensities on the number of electronic levels of the atoms | 18 pages, 11 figures, 8 tables, accepted for publication at MNRAS | null | 10.1093/mnras/staf605 | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | This paper presents a study of the dependence of the simulated intensities of
recombination lines from hydrogen and helium atoms on the number of
$n\ell$-resolved principal quantum numbers included in the calculations. We
simulate hydrogen and helium emitting astrophysical plasmas using the code
Cloudy and show that, if not enough $n\ell$-resolved levels are included,
recombination line intensities can be predicted with significant errors than
can be more than 30\% for H~I IR lines and 10\% for He~I optical lines
($\sim$20\% for He~I IR recombination lines) at densities
$\sim1\text{cm}^{-3}$, comparable to interstellar medium. This can have
consequences in several spectroscopic studies where high accuracy is required,
such as primordial helium abundance determination. Our results indicate that
the minimum number of resolved levels included in the simulated hydrogen and
helium ions of our spectral emission models should be adjusted to the specific
lines to be predicted, as well as to the temperature and density conditions of
the simulated plasma.
| [
{
"version": "v1",
"created": "Wed, 9 Apr 2025 17:38:40 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Guzmán",
"F.",
""
],
[
"Chatzikos",
"M.",
""
],
[
"Ferland",
"G.",
""
]
] |
2504.07822 | Wanna Cui | Wanna Cui and Peizheng Wang and Faliang Yin | DG-STMTL: A Novel Graph Convolutional Network for Multi-Task
Spatio-Temporal Traffic Forecasting | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/publicdomain/zero/1.0/ | Spatio-temporal traffic prediction is crucial in intelligent transportation
systems. The key challenge of accurate prediction is how to model the complex
spatio-temporal dependencies and adapt to the inherent dynamics in data.
Traditional Graph Convolutional Networks (GCNs) often struggle with static
adjacency matrices that introduce domain bias or learnable matrices that may be
overfitting to specific patterns. This challenge becomes more complex when
considering Multi-Task Learning (MTL). While MTL has the potential to enhance
prediction accuracy through task synergies, it can also face significant
hurdles due to task interference. To overcome these challenges, this study
introduces a novel MTL framework, Dynamic Group-wise Spatio-Temporal Multi-Task
Learning (DG-STMTL). DG-STMTL proposes a hybrid adjacency matrix generation
module that combines static matrices with dynamic ones through a task-specific
gating mechanism. We also introduce a group-wise GCN module to enhance the
modelling capability of spatio-temporal dependencies. We conduct extensive
experiments on two real-world datasets to evaluate our method. Results show
that our method outperforms other state-of-the-arts, indicating its
effectiveness and robustness.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 15:00:20 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 22:50:27 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Cui",
"Wanna",
""
],
[
"Wang",
"Peizheng",
""
],
[
"Yin",
"Faliang",
""
]
] |
2504.07948 | Jean-Philip Piquemal | Anouar Benali, Thomas Pl\'e, Olivier Adjoua, Valay Agarawal, Thomas
Applencourt, Marharyta Blazhynska, Raymond Clay III, Kevin Gasperich, Khalid
Hossain, Jeongnim Kim, Christopher Knight, Jaron T. Krogel, Yvon Maday,
Maxime Maria, Matthieu Montes, Ye Luo, Evgeny Posenitskiy, Corentin Villot,
Venkatram Vishwanath, Louis Lagard\`ere, Jean-Philip Piquemal | Pushing the Accuracy Limit of Foundation Neural Network Models with
Quantum Monte Carlo Forces and Path Integrals | null | null | null | null | physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | We propose an end-to-end integrated strategy to produce highly accurate
quantum chemistry (QC) synthetic datasets (energies and forces) aimed at
deriving Foundation Machine Learning models for molecular simulation. Starting
from Density Functional Theory (DFT), a "Jacob's Ladder" approach leverages
computationally-optimized layers of massively GPU-accelerated software with
increasing accuracy. Thanks to Exascale, this is the first time that the
computationally intensive calculation of Quantum Monte Carlo forces (QMC), and
the combination of multi-determinant QMC energies and forces with
selected-Configuration Interaction wavefunctions, are computed at such scale at
the complete basis-set limit. To bridge the gap between accurate QC and
condensed-phase Molecular Dynamics, we leverage transfer learning to improve
the DFT-based FeNNix-Bio1 foundation model. The resulting approach is coupled
to path integrals adaptive sampling quantum dynamics to perform nanosecond
reactive simulations at unprecedented accuracy. These results demonstrate the
promise of Exascale to deepen our understanding of the inner machinery of
complex biosystems.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 17:55:09 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 17:57:03 GMT"
},
{
"version": "v3",
"created": "Mon, 14 Apr 2025 17:06:38 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Benali",
"Anouar",
""
],
[
"Plé",
"Thomas",
""
],
[
"Adjoua",
"Olivier",
""
],
[
"Agarawal",
"Valay",
""
],
[
"Applencourt",
"Thomas",
""
],
[
"Blazhynska",
"Marharyta",
""
],
[
"Clay",
"Raymond",
"III"
],
[
"Gasperich",
"Kevin",
""
],
[
"Hossain",
"Khalid",
""
],
[
"Kim",
"Jeongnim",
""
],
[
"Knight",
"Christopher",
""
],
[
"Krogel",
"Jaron T.",
""
],
[
"Maday",
"Yvon",
""
],
[
"Maria",
"Maxime",
""
],
[
"Montes",
"Matthieu",
""
],
[
"Luo",
"Ye",
""
],
[
"Posenitskiy",
"Evgeny",
""
],
[
"Villot",
"Corentin",
""
],
[
"Vishwanath",
"Venkatram",
""
],
[
"Lagardère",
"Louis",
""
],
[
"Piquemal",
"Jean-Philip",
""
]
] |
2504.07983 | Shurui Wu | Shurui Wu, Xinyi Huang, Dingxin Lu | Psychological Health Knowledge-Enhanced LLM-based Social Network Crisis
Intervention Text Transfer Recognition Method | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | As the prevalence of mental health crises increases on social media
platforms, identifying and preventing potential harm has become an urgent
challenge. This study introduces a large language model (LLM)-based text
transfer recognition method for social network crisis intervention, enhanced
with domain-specific mental health knowledge. We propose a multi-level
framework that incorporates transfer learning using BERT, and integrates mental
health knowledge, sentiment analysis, and behavior prediction techniques. The
framework includes a crisis annotation tool trained on social media datasets
from real-world events, enabling the model to detect nuanced emotional cues and
identify psychological crises. Experimental results show that the proposed
method outperforms traditional models in crisis detection accuracy and exhibits
greater sensitivity to subtle emotional and contextual variations.
| [
{
"version": "v1",
"created": "Sat, 5 Apr 2025 22:57:22 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 01:47:33 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Wu",
"Shurui",
""
],
[
"Huang",
"Xinyi",
""
],
[
"Lu",
"Dingxin",
""
]
] |
2504.07993 | Viktor Ivanov | Viktor Ivanov, Richard C. Wilson, Maurizio Scaramuzza | Towards Simple Machine Learning Baselines for GNSS RFI Detection | null | null | null | null | eess.SP cs.LG | http://creativecommons.org/licenses/by/4.0/ | Machine learning research in GNSS radio frequency interference (RFI)
detection often lacks a clear empirical justification for the choice of deep
learning architectures over simpler machine learning approaches. In this work,
we argue for a change in research direction-from developing ever more complex
deep learning models to carefully assessing their real-world effectiveness in
comparison to interpretable and lightweight machine learning baselines. Our
findings reveal that state-of-the-art deep learning models frequently fail to
outperform simple, well-engineered machine learning methods in the context of
GNSS RFI detection. Leveraging a unique large-scale dataset collected by the
Swiss Air Force and Swiss Air-Rescue (Rega), and preprocessed by Swiss Air
Navigation Services Ltd. (Skyguide), we demonstrate that a simple baseline
model achieves 91\% accuracy in detecting GNSS RFI, outperforming more complex
deep learning counterparts. These results highlight the effectiveness of
pragmatic solutions and offer valuable insights to guide future research in
this critical application domain.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2025 12:45:01 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 06:59:33 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ivanov",
"Viktor",
""
],
[
"Wilson",
"Richard C.",
""
],
[
"Scaramuzza",
"Maurizio",
""
]
] |
2504.08015 | Mayeul Arminjon | Mayeul Arminjon | Progress in evaluating a possible electromagnetic interaction energy in
a gravitational field | 13 pages, including 6 figures. Text of a talk given at IARD 2024: The
14th Biennial Conference on Classical and Quantum Relativistic Dynamics of
Particles and Fields, Jun 2024, Helsinki Aalto University, Finland. V2:
abstract now complete: the end had been omitted by mistake | Journal of Physics: Conference Series 2987 (2025) 012006 | 10.1088/1742-6596/2987/1/012006 | null | physics.gen-ph | http://creativecommons.org/licenses/by/4.0/ | The Lorentz-Poincar\'e interpretation of special relativity (SR) keeps the
classical concepts of separated space and time, at the price of postulating an
indetectable preferred inertial frame or ``ether". But SR does not contain
gravity. The presence of gravity could make the ether detectable. This is one
idea behind the ``scalar ether theory of gravitation" (SET), which coincides
with SR if the gravity field vanishes, and passes a number of tests. However,
the coupling of SET with the Maxwell electromagnetic (EM) field needs to use
the theory's dynamical equation for the energy tensor in a non-trivial way. It
cannot be assumed that the energy tensors of the charged matter and the EM
field add to give the total energy tensor, source of the gravitational field.
Thus, an additional, ``interaction" energy tensor ${\bf T}_\mathrm{inter}$ has
to be postulated. Asking that ${\bf T}_\mathrm{inter}$ is Lorentz-invariant in
the situation of SR, fixes its form. It depends only on a scalar field $p$.
${\bf T}_\mathrm{inter}$ is an exotic kind of matter and is distributed in the
whole space, hence it could contribute to dark matter. For a weak gravitational
field, $p$ obeys a first-order partial differential equation (PDE) involving
the EM field and the Newtonian potential. However, the EM field varies on the
scale of the wavelength, which is extremely small. To get the field $p$ in a
galaxy, some averaging has to be done. After several attempts based on the
homogenization theory, a simpler way has been found recently: If the
macro-averages of $p$ and the EM field vary smoothly, it can be shown that the
PDE for $p$ remains valid in the same form with spacetime-averaged fields. The
current stage of calculations will also been shown.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 14:10:24 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 08:32:09 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Arminjon",
"Mayeul",
""
]
] |
2504.08016 | Soo Yong Lee | Soo Yong Lee, Hyunjin Hwang, Taekwan Kim, Yuyeong Kim, Kyuri Park,
Jaemin Yoo, Denny Borsboom, Kijung Shin | Emergence of psychopathological computations in large language models | pre-print | null | null | null | q-bio.NC cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Can large language models (LLMs) implement computations of psychopathology?
An effective approach to the question hinges on addressing two factors. First,
for conceptual validity, we require a general and computational account of
psychopathology that is applicable to computational entities without biological
embodiment or subjective experience. Second, mechanisms underlying LLM
behaviors need to be studied for better methodological validity. Thus, we
establish a computational-theoretical framework to provide an account of
psychopathology applicable to LLMs. To ground the theory for empirical
analysis, we also propose a novel mechanistic interpretability method alongside
a tailored empirical analytic framework. Based on the frameworks, we conduct
experiments demonstrating three key claims: first, that distinct dysfunctional
and problematic representational states are implemented in LLMs; second, that
their activations can spread and self-sustain to trap LLMs; and third, that
dynamic, cyclic structural causal models encoded in the LLMs underpin these
patterns. In concert, the empirical results corroborate our hypothesis that
network-theoretic computations of psychopathology have already emerged in LLMs.
This suggests that certain LLM behaviors mirroring psychopathology may not be a
superficial mimicry but a feature of their internal processing. Thus, our work
alludes to the possibility of AI systems with psychopathological behaviors in
the near future.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 15:36:30 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Lee",
"Soo Yong",
""
],
[
"Hwang",
"Hyunjin",
""
],
[
"Kim",
"Taekwan",
""
],
[
"Kim",
"Yuyeong",
""
],
[
"Park",
"Kyuri",
""
],
[
"Yoo",
"Jaemin",
""
],
[
"Borsboom",
"Denny",
""
],
[
"Shin",
"Kijung",
""
]
] |
2504.08045 | Zehao Jin | Zehao Jin and Joseph D. Gelfand | ZTF IC 10 variable catalog | 23 pages, 26 figures, 13 tables, accepted for publication in ApJS | null | 10.3847/1538-4365/adc7fe | null | astro-ph.GA astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | To study how massive variable stars effect their environment, we search for
variability among Zwicky Transient Facility (ZTF) sources located within the
optical extent of a nearby starburst galaxy IC 10. We present the ZTF IC 10
catalog, which classifies 1516 $r$ band sources and 864 $g$ band sources within
a $225''$ radius around IC 10 into three categories: 1388 (767) $r$ ($g$) band
non-variables, 150 (85) $r$ ($g$) band non-periodic variables, and 37 (12) $r$
($g$) band periodic variables. Among them 101 (48) $r$ ($g$) band non-periodic
variables, and 22 (4) $r$ ($g$) band periodic variables are inside IC 10. We
verify our classification by cross-matching with previous variability catalogs
and machine learning powered classifications. Various analysis including
population demographics, color-magnitude diagrams, and cross matching with a
set of different surveys and database such as Gaia, XMM-Newton, Chandra, and
SIMBAD are also presented. Based on source density and parallax, we distinguish
sources within IC 10 from non-IC 10 sources. For IC 10 sources, we highlight
flaring super giants, a source with long secondary period, periodic super
giants including possible S Doradus luminous blue variable and candidate Miras.
For non-IC 10 sources, we present super red sources and compact objects such as
a possible long period subdwarf and a periodic X-ray source. The catalog can
serve as a useful database to study the connection between various type of
massive stars and their host galaxies.
| [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 18:03:22 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 11:32:53 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Jin",
"Zehao",
""
],
[
"Gelfand",
"Joseph D.",
""
]
] |
2504.08248 | Mengfei He | Mengfei He, Sungkyu Cho, Gianna Dafflisio, Sitaram Emani, L. Mahadevan | Stochastic elastohydrodynamics of soft valves | Corrected typographical errors in figure and equation references
between the main text and supplementary information | null | null | null | cond-mat.soft physics.flu-dyn | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Soft valves serve to modulate and rectify flows in complex vasculatures
across the tree of life, e.g. in the heart of every human reading this. Here we
consider a minimal physical model of the heart mitral valve modeled as a
flexible conical shell capable of flow rectification via collapse and
coaptation in an impinging (reverse) flow. Our experiments show that the
complex elastohydrodynamics of closure features a noise-activated rectification
mechanism. A minimal theoretical model allows us to rationalize our
observations while illuminating a dynamical bifurcation driven by stochastic
hydrodynamic forces. Our theory also suggests a way to trigger the coaptation
of soft valves on demand, which we corroborate using experiments, suggesting a
design principle for their efficient operation.
| [
{
"version": "v1",
"created": "Fri, 11 Apr 2025 04:14:54 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 07:15:39 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"He",
"Mengfei",
""
],
[
"Cho",
"Sungkyu",
""
],
[
"Dafflisio",
"Gianna",
""
],
[
"Emani",
"Sitaram",
""
],
[
"Mahadevan",
"L.",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.