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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2504.09248 | Shuai Feng | Shuai Feng, Qian Ma, Junsoo Kim, Shengyuan Xu | Asymptotic stabilization under homomorphic encryption: A re-encryption
free method | null | null | null | null | eess.SY cs.CR cs.SY | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this paper, we propose methods to encrypted a pre-given dynamic controller
with homomorphic encryption, without re-encrypting the control inputs. We first
present a preliminary result showing that the coefficients in a pre-given
dynamic controller can be scaled up into integers by the zooming-in factor in
dynamic quantization, without utilizing re-encryption. However, a sufficiently
small zooming-in factor may not always exist because it requires that the
convergence speed of the pre-given closed-loop system should be sufficiently
fast. Then, as the main result, we design a new controller approximating the
pre-given dynamic controller, in which the zooming-in factor is decoupled from
the convergence rate of the pre-given closed-loop system. Therefore, there
always exist a (sufficiently small) zooming-in factor of dynamic quantization
scaling up all the controller's coefficients to integers, and a finite modulus
preventing overflow in cryptosystems. The process is asymptotically stable and
the quantizer is not saturated.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:07:34 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Feng",
"Shuai",
""
],
[
"Ma",
"Qian",
""
],
[
"Kim",
"Junsoo",
""
],
[
"Xu",
"Shengyuan",
""
]
] |
2504.09249 | Soumitri Chattopadhyay | Aniket Pal, Sanket Biswas, Alloy Das, Ayush Lodh, Priyanka Banerjee,
Soumitri Chattopadhyay, Dimosthenis Karatzas, Josep Llados, C.V. Jawahar | NoTeS-Bank: Benchmarking Neural Transcription and Search for Scientific
Notes Understanding | null | null | null | null | cs.CV cs.IR cs.LG cs.MM | http://creativecommons.org/licenses/by/4.0/ | Understanding and reasoning over academic handwritten notes remains a
challenge in document AI, particularly for mathematical equations, diagrams,
and scientific notations. Existing visual question answering (VQA) benchmarks
focus on printed or structured handwritten text, limiting generalization to
real-world note-taking. To address this, we introduce NoTeS-Bank, an evaluation
benchmark for Neural Transcription and Search in note-based question answering.
NoTeS-Bank comprises complex notes across multiple domains, requiring models to
process unstructured and multimodal content. The benchmark defines two tasks:
(1) Evidence-Based VQA, where models retrieve localized answers with
bounding-box evidence, and (2) Open-Domain VQA, where models classify the
domain before retrieving relevant documents and answers. Unlike classical
Document VQA datasets relying on optical character recognition (OCR) and
structured data, NoTeS-BANK demands vision-language fusion, retrieval, and
multimodal reasoning. We benchmark state-of-the-art Vision-Language Models
(VLMs) and retrieval frameworks, exposing structured transcription and
reasoning limitations. NoTeS-Bank provides a rigorous evaluation with NDCG@5,
MRR, Recall@K, IoU, and ANLS, establishing a new standard for visual document
understanding and reasoning.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:11:32 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Pal",
"Aniket",
""
],
[
"Biswas",
"Sanket",
""
],
[
"Das",
"Alloy",
""
],
[
"Lodh",
"Ayush",
""
],
[
"Banerjee",
"Priyanka",
""
],
[
"Chattopadhyay",
"Soumitri",
""
],
[
"Karatzas",
"Dimosthenis",
""
],
[
"Llados",
"Josep",
""
],
[
"Jawahar",
"C. V.",
""
]
] |
2504.09250 | Keisuke Murota | Keisuke Murota | Adiabatic Encoding of Pre-trained MPS Classifiers into Quantum Circuits | null | null | null | null | quant-ph cond-mat.str-el cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Although Quantum Neural Networks (QNNs) offer powerful methods for
classification tasks, the training of QNNs faces two major training obstacles:
barren plateaus and local minima. A promising solution is to first train a
tensor-network (TN) model classically and then embed it into a QNN.\ However,
embedding TN-classifiers into quantum circuits generally requires postselection
whose success probability may decay exponentially with the system size. We
propose an \emph{adiabatic encoding} framework that encodes pre-trained
MPS-classifiers into quantum MPS (qMPS) circuits with postselection, and
gradually removes the postselection while retaining performance. We prove that
training qMPS-classifiers from scratch on a certain artificial dataset is
exponentially hard due to barren plateaus, but our adiabatic encoding
circumvents this issue. Additional numerical experiments on binary MNIST also
confirm its robustness.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:12:46 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Murota",
"Keisuke",
""
]
] |
2504.09251 | Xiaxing Cai | Xiaxing Cai | Affine Logarithmic HLS and Beckner-Type Logarithmic Sobolev Inequalities | null | null | null | null | math.MG math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we consider two limiting cases ($\alpha\rightarrow n$ and
$\alpha\rightarrow 0 $) of the recent affine HLS inequalities by Haddad and
Ludwig. As $\alpha\rightarrow n$, the affine logarithmic HLS inequality is
established, which is stronger than the logarithmic HLS inequality by Carlen
and Loss from 1992 and Beckner from 1993. As $\alpha\rightarrow 0$, an affine
version of Beckner's logarithmic Sobolev inequality is established, which is
also a limiting case of the affine fractional $L^2$ Sobolev inequalities. The
affine logarithmic Sobolev inequality is stronger than the original version by
Beckner from 1995.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:19:33 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Cai",
"Xiaxing",
""
]
] |
2504.09252 | Shou-Fu Tian | Jia-Fu Tong and Shou-Fu Tian | On Cauchy problem to the modified Camassa-Holm equation: Painlev\'{e}
asymptotics | 56 pages, 12 figures | null | null | null | math.AP math-ph math.MP nlin.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the Painlev\'{e} asymptotics for the Cauchy problem of the
modified Camassa-Holm (mCH) equation with zero boundary conditions
\begin{align*}\nonumber &m_t+\left((u^2-u_x^2)m\right)_x=0, \
(x,t)\in\mathbb{R}\times\mathbb{R}^+,\\ &u(x,0)=u_0(x), \lim_{x\to\pm\infty}
u_0(x)=0, \end{align*} where $u_0(x)\in H^{4,2}(\mathbb{R})$. Recently, Yang
and Fan (Adv. Math. 402, 108340 (2022)) reported the long-time asymptotic
result for the mCH equation in the solitonic regions. The main purpose of our
work is to study the long-time asymptotic behavior in two transition regions.
The key to proving this result is to establish and analyze the Riemann-Hilbert
problem on a new plane $(y;t)$ related to the Cauchy problem of the mCH
equation. With the $\bar{\partial}$-generalization of the Deift-Zhou nonlinear
steepest descent method and double scaling limit technique, in two transition
regions defined by \begin{align}\nonumber \mathcal{P}_{I}:=\{(x,t):0\leqslant
|\frac{x}{t}-2|t^{2/3}\leqslant C\},~~~~\mathcal{P}_{II}:=\{(x,t):0\leqslant
|\frac{x}{t}+1/4|t^{2/3}\leqslant C\}, \end{align} where $C>0$ is a constant,
we find that the leading order approximation to the solution of the mCH
equation can be expressed in terms of the Painlev\'{e} II equation.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:20:04 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Tong",
"Jia-Fu",
""
],
[
"Tian",
"Shou-Fu",
""
]
] |
2504.09253 | Dian Zheng | Dian Zheng, Lingzhou Xue | Statistical Inference for High-Dimensional Robust Linear Regression
Models via Recursive Online-Score Estimation | null | null | null | null | stat.ME math.ST stat.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper introduces a novel framework for estimation and inference in
penalized M-estimators applied to robust high-dimensional linear regression
models. Traditional methods for high-dimensional statistical inference, which
predominantly rely on convex likelihood-based approaches, struggle to address
the nonconvexity inherent in penalized M-estimation with nonconvex objective
functions. Our proposed method extends the recursive online score estimation
(ROSE) framework of Shi et al. (2021) to robust high-dimensional settings by
developing a recursive score equation based on penalized M-estimation,
explicitly addressing nonconvexity. We establish the statistical consistency
and asymptotic normality of the resulting estimator, providing a rigorous
foundation for valid inference in robust high-dimensional regression. The
effectiveness of our method is demonstrated through simulation studies and a
real-world application, showcasing its superior performance compared to
existing approaches.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:22:21 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zheng",
"Dian",
""
],
[
"Xue",
"Lingzhou",
""
]
] |
2504.09255 | Sijing Wu | Sijing Wu, Yunhao Li, Ziwen Xu, Yixuan Gao, Huiyu Duan, Wei Sun,
Guangtao Zhai | FVQ: A Large-Scale Dataset and A LMM-based Method for Face Video Quality
Assessment | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Face video quality assessment (FVQA) deserves to be explored in addition to
general video quality assessment (VQA), as face videos are the primary content
on social media platforms and human visual system (HVS) is particularly
sensitive to human faces. However, FVQA is rarely explored due to the lack of
large-scale FVQA datasets. To fill this gap, we present the first large-scale
in-the-wild FVQA dataset, FVQ-20K, which contains 20,000 in-the-wild face
videos together with corresponding mean opinion score (MOS) annotations. Along
with the FVQ-20K dataset, we further propose a specialized FVQA method named
FVQ-Rater to achieve human-like rating and scoring for face video, which is the
first attempt to explore the potential of large multimodal models (LMMs) for
the FVQA task. Concretely, we elaborately extract multi-dimensional features
including spatial features, temporal features, and face-specific features
(i.e., portrait features and face embeddings) to provide comprehensive visual
information, and take advantage of the LoRA-based instruction tuning technique
to achieve quality-specific fine-tuning, which shows superior performance on
both FVQ-20K and CFVQA datasets. Extensive experiments and comprehensive
analysis demonstrate the significant potential of the FVQ-20K dataset and
FVQ-Rater method in promoting the development of FVQA.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:26:02 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Wu",
"Sijing",
""
],
[
"Li",
"Yunhao",
""
],
[
"Xu",
"Ziwen",
""
],
[
"Gao",
"Yixuan",
""
],
[
"Duan",
"Huiyu",
""
],
[
"Sun",
"Wei",
""
],
[
"Zhai",
"Guangtao",
""
]
] |
2504.09256 | Mohamad Nasser | Mohamad N. Nasser | On extensions of the standard representation of the braid group to the
singular braid group | null | null | null | null | math.RT | http://creativecommons.org/licenses/by/4.0/ | For an integer $n \geq 2$, set $B_n$ to be the braid group on $n$ strands and
$SB_n$ to be the singular braid group on $n$ strands. $SB_n$ is one of the
important group extensions of $B_n$ that appeared in 1998. Our aim in this
paper is to extend the well-known standard representation of $B_n$, namely
$\rho_S:B_n \to GL_n(\mathbb{Z}[t^{\pm 1}])$, to $SB_n$, for all $n \geq 2$,
and to investigate the characteristics of these extended representations as
well. The first major finding in our paper is that we determine the form of all
representations of $SB_n$, namely $\rho'_S: SB_n \to GL_n(\mathbb{Z}[t^{\pm
1}])$, that extend $\rho_S$, for all $n\geq 2$. The second major finding is
that we find necessary and sufficient conditions for the irreduciblity of the
representations of the form $\rho'_S$ of $SB_n$, for all $n\geq 2$. We prove
that, for $t\neq 1$, the representations of the form $\rho'_S$ are irreducible
and, for $t=1$, the representations of the form $\rho'_S$ are irreducible if
and only if $a+c\neq 1.$ The third major result is that we consider the virtual
singular braid group on $n$ strands, $VSB_n$, which is a group extension of
both $B_n$ and $SB_n$, and we determine the form of all representations
$\rho''_S: VSB_2 \to GL_2(\mathbb{Z}[t^{\pm 1}])$, that extend $\rho_S$ and
$\rho'_S$; making a path toward finding the form of all representations
$\rho''_S: VSB_n \to GL_n(\mathbb{Z}[t^{\pm 1}])$, that extend $\rho_S$ and
$\rho'_S$, for all $n\geq 3$.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:30:44 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Nasser",
"Mohamad N.",
""
]
] |
2504.09257 | Sohom Ghosh | Sohom Ghosh and Arnab Maji and Sudip Kumar Naskar | MiMIC: Multi-Modal Indian Earnings Calls Dataset to Predict Stock Prices | Code and Dataset:
https://huggingface.co/datasets/sohomghosh/MiMIC_Multi-Modal_Indian_Earnings_Calls_Dataset | null | null | null | cs.LG cs.AI cs.CL | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Predicting stock market prices following corporate earnings calls remains a
significant challenge for investors and researchers alike, requiring innovative
approaches that can process diverse information sources. This study
investigates the impact of corporate earnings calls on stock prices by
introducing a multi-modal predictive model. We leverage textual data from
earnings call transcripts, along with images and tables from accompanying
presentations, to forecast stock price movements on the trading day immediately
following these calls. To facilitate this research, we developed the MiMIC
(Multi-Modal Indian Earnings Calls) dataset, encompassing companies
representing the Nifty 50, Nifty MidCap 50, and Nifty Small 50 indices. The
dataset includes earnings call transcripts, presentations, fundamentals,
technical indicators, and subsequent stock prices. We present a multimodal
analytical framework that integrates quantitative variables with predictive
signals derived from textual and visual modalities, thereby enabling a holistic
approach to feature representation and analysis. This multi-modal approach
demonstrates the potential for integrating diverse information sources to
enhance financial forecasting accuracy. To promote further research in
computational economics, we have made the MiMIC dataset publicly available
under the CC-NC-SA-4.0 licence. Our work contributes to the growing body of
literature on market reactions to corporate communications and highlights the
efficacy of multi-modal machine learning techniques in financial analysis.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:31:40 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ghosh",
"Sohom",
""
],
[
"Maji",
"Arnab",
""
],
[
"Naskar",
"Sudip Kumar",
""
]
] |
2504.09259 | Claudia Filippi | Stuart Shepard, Anthony Scemama, Saverio Moroni, and Claudia Filippi | Optimizing excited states in quantum Monte Carlo: A reassessment of
double excitations | 16 pages, 2 figures | null | null | null | physics.comp-ph physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | Quantum Monte Carlo (QMC) methods have proven to be highly accurate for
computing excited states, but the choice of optimization strategies for
multiple states remains an active topic of investigation. In this work, we
revisit the calculation of double excitation energies in nitroxyl, glyoxal,
tetrazine, and cyclopentadienone, exploring different objective functionals and
their impact on the accuracy and robustness of QMC. A previous study for these
systems employed a penalty functional to enforce orthogonality among the
states, but the chosen prefactors did not strictly ensure convergence to the
target states. Here, we confirm the reliability of previous results by
comparing excitation energies obtained with different functionals and analyzing
their consistency. Additionally, we investigate the performance of different
functionals when starting from a pre-collapsed excited state, providing insight
into their ability to recover the target wave functions.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:35:25 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Shepard",
"Stuart",
""
],
[
"Scemama",
"Anthony",
""
],
[
"Moroni",
"Saverio",
""
],
[
"Filippi",
"Claudia",
""
]
] |
2504.09260 | Wenji Fang | Wenji Fang, Wenkai Li, Shang Liu, Yao Lu, Hongce Zhang, Zhiyao Xie | NetTAG: A Multimodal RTL-and-Layout-Aligned Netlist Foundation Model via
Text-Attributed Graph | Accepted by Design Automation Conference (DAC), 2025 | null | null | null | cs.AR cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Circuit representation learning has shown promise in advancing Electronic
Design Automation (EDA) by capturing structural and functional circuit
properties for various tasks. Existing pre-trained solutions rely on graph
learning with complex functional supervision, such as truth table simulation.
However, they only handle simple and-inverter graphs (AIGs), struggling to
fully encode other complex gate functionalities. While large language models
(LLMs) excel at functional understanding, they lack the structural awareness
for flattened netlists. To advance netlist representation learning, we present
NetTAG, a netlist foundation model that fuses gate semantics with graph
structure, handling diverse gate types and supporting a variety of functional
and physical tasks. Moving beyond existing graph-only methods, NetTAG
formulates netlists as text-attributed graphs, with gates annotated by symbolic
logic expressions and physical characteristics as text attributes. Its
multimodal architecture combines an LLM-based text encoder for gate semantics
and a graph transformer for global structure. Pre-trained with gate and graph
self-supervised objectives and aligned with RTL and layout stages, NetTAG
captures comprehensive circuit intrinsics. Experimental results show that
NetTAG consistently outperforms each task-specific method on four largely
different functional and physical tasks and surpasses state-of-the-art AIG
encoders, demonstrating its versatility.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:39:25 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Fang",
"Wenji",
""
],
[
"Li",
"Wenkai",
""
],
[
"Liu",
"Shang",
""
],
[
"Lu",
"Yao",
""
],
[
"Zhang",
"Hongce",
""
],
[
"Xie",
"Zhiyao",
""
]
] |
2504.09261 | Ziran Qin | Ziran Qin, Youru Lv, Mingbao Lin, Zeren Zhang, Danping Zou, Weiyao Lin | Head-Aware KV Cache Compression for Efficient Visual Autoregressive
Modeling | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Visual Autoregressive (VAR) models have emerged as a powerful approach for
multi-modal content creation, offering high efficiency and quality across
diverse multimedia applications. However, they face significant memory
bottlenecks due to extensive KV cache accumulation during inference. Existing
KV cache compression techniques for large language models are suboptimal for
VAR models due to, as we identify in this paper, two distinct categories of
attention heads in VAR models: Structural Heads, which preserve spatial
coherence through diagonal attention patterns, and Contextual Heads, which
maintain semantic consistency through vertical attention patterns. These
differences render single-strategy KV compression techniques ineffective for
VAR models. To address this, we propose HACK, a training-free Head-Aware
Compression method for KV cache. HACK allocates asymmetric cache budgets and
employs pattern-specific compression strategies tailored to the essential
characteristics of each head category. Experiments on Infinity-2B, Infinity-8B,
and VAR-d30 demonstrate its effectiveness in text-to-image and
class-conditional generation tasks. HACK can hack down up to 50\% and 70\% of
cache with minimal performance degradation for VAR-d30 and Infinity-8B,
respectively. Even with 70\% and 90\% KV cache compression in VAR-d30 and
Infinity-8B, HACK still maintains high-quality generation while reducing memory
usage by 44.2\% and 58.9\%, respectively.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:42:17 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Qin",
"Ziran",
""
],
[
"Lv",
"Youru",
""
],
[
"Lin",
"Mingbao",
""
],
[
"Zhang",
"Zeren",
""
],
[
"Zou",
"Danping",
""
],
[
"Lin",
"Weiyao",
""
]
] |
2504.09262 | Qi Lu | Zengyu Li, Zhonghua Liao and Qi L\"u | Exact Controllability for a Refined Stochastic Hyperbolic Equation with
Internal Controls | null | null | null | null | math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We establish the internal exact controllability of a refined stochastic
hyperbolic equation by deriving a suitable observability inequality via
Carleman estimates for the associated backward stochastic hyperbolic equation.
In contrast to existing results on boundary exact controllability--which
require longer waiting times, we demonstrate that the required waiting time for
internal exact controllability in stochastic hyperbolic equations coincides
exactly with that of their deterministic counterparts.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:45:57 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Li",
"Zengyu",
""
],
[
"Liao",
"Zhonghua",
""
],
[
"Lü",
"Qi",
""
]
] |
2504.09263 | Rodrigo de Lamare | S. Salehi, S. Mashdour, O. Tamyigit, S. Seyedmasoumian, M. Moradikia,
R. C. de Lamare and A.Schmeink | Machine Learning-Based AP Selection in User-Centric Cell-free
Multiple-Antenna Networks | 7 pages, 5 figures | null | null | null | cs.IT math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | User-centric cell-free (UCCF) massive multiple-input multiple-output (MIMO)
systems are considered a viable solution to realize the advantages offered by
cell-free (CF) networks, including reduced interference and consistent quality
of service while maintaining manageable complexity. In this paper, we propose
novel learning-based access point (AP) selection schemes tailored for UCCF
massive MIMO systems. The learning model exploits the dataset generated from
two distinct AP selection schemes, based on large-scale fading (LSF)
coefficients and the sum-rate coefficients, respectively. The proposed
learning-based AP selection schemes could be implemented centralized or
distributed, with the aim of performing AP selection efficiently. We evaluate
our model's performance against CF and two heuristic clustering schemes for
UCCF networks. The results demonstrate that the learning-based approach
achieves a comparable sum-rate performance to that of competing techniques for
UCCF networks, while significantly reducing computational complexity.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:47:01 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Salehi",
"S.",
""
],
[
"Mashdour",
"S.",
""
],
[
"Tamyigit",
"O.",
""
],
[
"Seyedmasoumian",
"S.",
""
],
[
"Moradikia",
"M.",
""
],
[
"de Lamare",
"R. C.",
""
],
[
"Schmeink",
"A.",
""
]
] |
2504.09264 | Peter \v{S}vihra | Peter \v{S}vihra, Jan Chochol, Vladim\'ir Kafka, Adam Klimsza, Adam
Kozelsky, Ji\v{r}\'i Kroll, Roman Malousek, M\'aria Mar\v{c}i\v{s}ovsk\'a,
Michal Mar\v{c}i\v{s}ovsk\'y, Marcela Mike\v{s}t\'ikov\'a, Michael Moll,
David Nov\'ak, Radek Novotn\'y, Peter Slov\'ak, Radim \v{S}pet\'ik, Moritz
Wiehe | Exploring the Design and Measurements of Next-Generation 4H-SiC LGADs | null | null | null | null | physics.ins-det | http://creativecommons.org/licenses/by/4.0/ | This contribution presents the design, production, and initial testing of
newly developed 4H-SiC Low Gain Avalanche Detectors (LGADs). The evaluation
includes performance metrics such as the internal gain layer's efficiency in
enhancing signal generation. Initial laboratory and Transient Current Technique
(TCT) measurements provide insight into the device's stability and response to
the signal.
Due to the increase of availability provided by the industry, 4H-SiC is
emerging as a strong candidate for the next-generation of semiconductor
detectors. Such sensors are promising due to the inherent radiation tolerance
of 4H-SiC and its stable operation across a wide temperature range. However,
due to the wider-bandgap of 4H-SiC compared to standard silicon, and difficulty
to produce high-quality layers thicker than 50 \textmu m, an internal charge
multiplication layer needs to be introduced.
The presented 4H-SiC LGADs, fabricated by onsemi, are optimized for an N-type
substrate and epi wafer. The initial TCT and laboratory test results
demonstrate fast charge collection and uniform multiplication across multiple
samples produced on a single wafer.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:54:38 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Švihra",
"Peter",
""
],
[
"Chochol",
"Jan",
""
],
[
"Kafka",
"Vladimír",
""
],
[
"Klimsza",
"Adam",
""
],
[
"Kozelsky",
"Adam",
""
],
[
"Kroll",
"Jiří",
""
],
[
"Malousek",
"Roman",
""
],
[
"Marčišovská",
"Mária",
""
],
[
"Marčišovský",
"Michal",
""
],
[
"Mikeštíková",
"Marcela",
""
],
[
"Moll",
"Michael",
""
],
[
"Novák",
"David",
""
],
[
"Novotný",
"Radek",
""
],
[
"Slovák",
"Peter",
""
],
[
"Špetík",
"Radim",
""
],
[
"Wiehe",
"Moritz",
""
]
] |
2504.09265 | Lei Kang | Lei Kang, Jia Li, Mi Tian and Hua Huang | Mixture of Group Experts for Learning Invariant Representations | null | null | null | null | cs.LG cs.CL cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Sparsely activated Mixture-of-Experts (MoE) models effectively increase the
number of parameters while maintaining consistent computational costs per
token. However, vanilla MoE models often suffer from limited diversity and
specialization among experts, constraining their performance and scalability,
especially as the number of experts increases. In this paper, we present a
novel perspective on vanilla MoE with top-$k$ routing inspired by sparse
representation. This allows us to bridge established theoretical insights from
sparse representation into MoE models. Building on this foundation, we propose
a group sparse regularization approach for the input of top-$k$ routing, termed
Mixture of Group Experts (MoGE). MoGE indirectly regularizes experts by
imposing structural constraints on the routing inputs, while preserving the
original MoE architecture. Furthermore, we organize the routing input into a 2D
topographic map, spatially grouping neighboring elements. This structure
enables MoGE to capture representations invariant to minor transformations,
thereby significantly enhancing expert diversity and specialization.
Comprehensive evaluations across various Transformer models for image
classification and language modeling tasks demonstrate that MoGE substantially
outperforms its MoE counterpart, with minimal additional memory and computation
overhead. Our approach provides a simple yet effective solution to scale the
number of experts and reduce redundancy among them. The source code is included
in the supplementary material and will be publicly released.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 15:58:02 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Kang",
"Lei",
""
],
[
"Li",
"Jia",
""
],
[
"Tian",
"Mi",
""
],
[
"Huang",
"Hua",
""
]
] |
2504.09266 | Jameel-Un Nabi | Jameel-Un Nabi, Tuncay Bayram | Nuclear structure properties and decay rates of molybdenum isotopes | 24 Pages, 10 Figures, 4 Tables | Astrophysics and Space Science, 365(1), 1-16 (2020) | 10.1007/s10509-020-3735-5 | null | nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Electron capture and beta-minus decay are the dominant decay processes during
the late phases of the evolution of heavy stars. Previous simulation results
show that weak rates on isotopes of Molybdenum (Mo) have a meaningful
contribution during the development of phases of stars before they go
supernova. The relative abundance, coupled with the stellar weak rates on Mo
isotopes, may change the lepton-to-baryon content of the core material. Here,
we report on the calculation of nuclear structure properties of Mo isotopes
from mass number 82 to 138, employing the RMF model. Later, we calculate the
weak decay rates of these isotopes using the proton-neutron quasiparticle
random phase approximation (pn-QRPA) model. In the first step, the ground-state
nuclear properties of Mo isotopes such as binding energy per nucleon, neutron
and proton separation energies, charge radii, total electric quadrupole
moments, and the deformation parameter of electric quadrupole moments have been
calculated using the density-dependent version of the RMF model with DD-PC1 and
DD-ME2 functionals. The calculated electric quadrupole deformation parameters
have been used in a deformed pn-QRPA calculation in the second phase of this
work to calculate half-lives and weak decay rates for these Mo isotopes in
stellar matter. We calculate the electron capture and beta-decay rates over an
extensive range of temperature (0.01 x 10^9 K to 30 x 10^9 K) and density (10
to 10^11 g/cm^3). Our study can prove useful for simulation of presupernova
evolution processes of stars.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:05:45 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Nabi",
"Jameel-Un",
""
],
[
"Bayram",
"Tuncay",
""
]
] |
2504.09267 | Mengfei He | Mengfei He, Samay Hulikal, Marianna Marquardt, Hao Jiang, Anupam
Pandey, Teng Zhang, Christian D. Santangelo and Joseph D. Paulsen | Controlling Droplets at the Tips of Fibers | null | null | null | null | physics.flu-dyn cond-mat.soft | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Many complex wetting behaviors of fibrous materials are rooted in the
behaviors of individual droplets attached to pairs of fibers. Here, we study
the splitting of a droplet held between the tips of two cylindrical fibers. We
discover a sharp transition between two post-rupture states, navigated by
changing the angle between the rods, in agreement with our bifurcation
analysis. Depinning of the bridge contact line can lead to a much larger
asymmetry between the volume of liquid left on each rod. This second scenario
enables the near-complete transfer of an aqueous glycerol droplet between two
identical vinylpolysiloxane fibers. We leverage this response in a device that
uses a ruck to pass a droplet along a train of fibers, a proof-of-concept for
the geometric control of droplets on deformable, architected surfaces.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:05:52 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"He",
"Mengfei",
""
],
[
"Hulikal",
"Samay",
""
],
[
"Marquardt",
"Marianna",
""
],
[
"Jiang",
"Hao",
""
],
[
"Pandey",
"Anupam",
""
],
[
"Zhang",
"Teng",
""
],
[
"Santangelo",
"Christian D.",
""
],
[
"Paulsen",
"Joseph D.",
""
]
] |
2504.09268 | Rebekah Herrman | Mostafa Atallah and James Ostrowski and Rebekah Herrman | A mixed-integer program for circuit execution time minimization with
precedence constraints | null | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | We present a mixed-integer programming (MIP) model for scheduling quantum
circuits to minimize execution time. Our approach maximizes parallelism by
allowing non-overlapping gates (those acting on distinct qubits) to execute
simultaneously. Our methods apply to general circuits with precedence
constraints. First, we derive closed-formulas for the execution time of
circuits generated by ma-QAOA on star graphs for a layered, greedy, and MIP
schedules. We then compare the MIP schedule against layered and greedy
scheduling approaches on the circuits generated by ma-QAOA for solving the
MaxCut problem on all non-isomorphic connected graphs with 3-7 vertices. These
experiments demonstrate that the MIP scheduler consistently results in shorter
circuit execution times than greedy and layered approaches, with up to 24\%
savings.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:06:09 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Atallah",
"Mostafa",
""
],
[
"Ostrowski",
"James",
""
],
[
"Herrman",
"Rebekah",
""
]
] |
2504.09269 | Yann-Meing Law | Yann-Meing Law, Zhichao Peng, Daniel Appel\"o, Thomas Hagstrom | A $P$-Adaptive Hermite Method for Nonlinear Dispersive Maxwell's
Equations | 22 pages, 15 figures | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we introduce a novel Hermite method to handle Maxwell's
equations for nonlinear dispersive media. The proposed method achieves
high-order accuracy and is free of any nonlinear algebraic solver, requiring
solving instead small local linear systems for which the dimension is
independent of the order. The implementation of order adaptive algorithms is
straightforward in this setting, making the resulting p-adaptive Hermite method
appealing for the simulations of soliton-like wave propagation.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:09:07 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Law",
"Yann-Meing",
""
],
[
"Peng",
"Zhichao",
""
],
[
"Appelö",
"Daniel",
""
],
[
"Hagstrom",
"Thomas",
""
]
] |
2504.09270 | Yitong Wang | Yitong Wang | Diamond diagrams and multivariable
$(\varphi,\mathcal{O}_K^{\times})$-modules | 34 pages | null | null | null | math.NT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Let $p$ be a prime number and $K$ a finite unramified extension of
$\mathbb{Q}_p$. Let $\pi$ be an admissible smooth mod $p$ representation of
$\mathrm{GL}_2(K)$ occurring in some Hecke eigenspaces of the mod $p$
cohomology and $\overline{r}$ be its underlying global two-dimensional Galois
representation. When $\overline{r}$ satisfies some Taylor-Wiles hypotheses and
is sufficiently generic at $p$, we compute explicitly certain constants
appearing in the diagram associated to $\pi$, generalizing the results of
Dotto-Le. As a result, we prove that the associated \'etale
$(\varphi,\mathcal{O}_K^{\times})$-module $D_A(\pi)$ defined by
Breuil-Herzig-Hu-Morra-Schraen is explicitly determined by the restriction of
$\overline{r}$ to the decomposition group at $p$, generalizing the results of
Breuil-Herzig-Hu-Morra-Schraen and the author.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:14:47 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Wang",
"Yitong",
""
]
] |
2504.09271 | Koustuv Saha | Koustuv Saha, Yoshee Jain, and Munmun De Choudhury | Linguistic Comparison of AI- and Human-Written Responses to Online
Mental Health Queries | null | null | null | null | cs.HC cs.AI cs.CL cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The ubiquity and widespread use of digital and online technologies have
transformed mental health support, with online mental health communities
(OMHCs) providing safe spaces for peer support. More recently, generative AI
and large language models (LLMs) have introduced new possibilities for
scalable, around-the-clock mental health assistance that could potentially
augment and supplement the capabilities of OMHCs. Although genAI shows promise
in delivering immediate and personalized responses, their effectiveness in
replicating the nuanced, experience-based support of human peers remains an
open question. In this study, we harnessed 24,114 posts and 138,758 online
community (OC) responses from 55 OMHCs on Reddit. We prompted several
state-of-the-art LLMs (GPT-4-Turbo, Llama-3, and Mistral-7B) with these posts,
and compared their (AI) responses to human-written (OC) responses based on a
variety of linguistic measures across psycholinguistics and lexico-semantics.
Our findings revealed that AI responses are more verbose, readable, and
analytically structured, but lack linguistic diversity and personal narratives
inherent in human-human interactions. Through a qualitative examination, we
found validation as well as complementary insights into the nature of AI
responses, such as its neutrality of stance and the absence of seeking
back-and-forth clarifications. We discuss the ethical and practical
implications of integrating generative AI into OMHCs, advocating for frameworks
that balance AI's scalability and timeliness with the irreplaceable
authenticity, social interactiveness, and expertise of human connections that
form the ethos of online support communities.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:20:02 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Saha",
"Koustuv",
""
],
[
"Jain",
"Yoshee",
""
],
[
"De Choudhury",
"Munmun",
""
]
] |
2504.09272 | Juan Carlos De los Reyes | Juan Carlos De Los Reyes | Bouligand Analysis and Discrete Optimal Control of Total Variation-Based
Variational Inequalities | null | null | null | null | math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate differentiability and subdifferentiability properties of the
solution mapping associated with variational inequalities (VI) of the second
kind involving the discrete total-variation. Bouligand differentiability of the
solution operator is established via a direct quotient analysis applied to a
primal-dual reformulation of the VI. By exploiting the structure of the
directional derivative and introducing a suitable subspace, we fully
characterize the Bouligand subdifferential of the solution mapping. We then
derive optimality conditions characterizing Bouligand-stationary and
strongly-stationary points for discrete VI-constrained optimal control
problems. A trust-region algorithm for solving these control problems is
proposed based on the obtained characterizations, and a numerical experiment is
presented to illustrate the main properties of both the solution and the
proposed algorithm.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:22:30 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Reyes",
"Juan Carlos De Los",
""
]
] |
2504.09274 | Davide Barilari | Davide Barilari, Tania Bossio, Valentina Franceschi | Magnetic fields on sub-Riemannian manifolds | 27 pages | null | null | null | math.DG math.DS | http://creativecommons.org/licenses/by/4.0/ | Motivated by a classical correspondence between magnetic flows and
sub-Riemannian geometry, first established by Montgomery, we undertake a
systematic study of magnetic flows on sub-Riemannian manifolds.
We focus on three-dimensional contact manifolds, and we show that magnetic
fields are naturally defined through Rumin differential forms. We provide a
geometric interpretation of the sub-Riemannian magnetic geodesic flow,
demonstrating that it can be understood as a geodesic flow on a suitably
defined lifted sub-Riemannian structure, which is of Engel type when the
magnetic field is non-vanishing.
In the general case, when the magnetic field might be vanishing, we
investigate the geometry of this lifted structure, characterizing properties
such as its step and the abnormal trajectories in terms of the analytical
features of the magnetic field.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:43:49 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Barilari",
"Davide",
""
],
[
"Bossio",
"Tania",
""
],
[
"Franceschi",
"Valentina",
""
]
] |
2504.09275 | Raghunath Sahoo | Sarthak Tripathy, Suraj Prasad, and Raghunath Sahoo | Event shape dependence of symmetry plane correlations using the Gaussian
estimator in Pb-Pb collisions at the LHC using AMPT | 12 pages, 7 captioned figures. Submitted for publication | null | null | null | nucl-ex hep-ex hep-ph hep-th nucl-th | http://creativecommons.org/licenses/by-sa/4.0/ | The study of symmetry plane correlations (SPCs) can be useful in
characterizing the direction of the anisotropic emission of produced particles
in the final state. The study of SPCs provides an independent method to
understand the transport properties of the system formed in heavy-ion
collisions. Similar to anisotropic flow coefficients, which are largely
influenced by the initial spatial anisotropy, SPCs also depend upon the
participant plane correlation measured using the participating nucleons of the
collision overlap region. In this paper, SPCs have been studied in Pb-Pb
collisions at $\sqrt{s_{\rm NN}}=5.02$ TeV using the event generator AMPT. In
addition to their behaviour with the changing centrality of the collision,
their event shape dependence has also been studied for the first time, using
the event shape classifier transverse spherocity. The Gaussian estimator has
been used to evaluate the correlations, and these have been compared to the
participant plane correlations defined in an analogous way to the symmetry
plane correlations and a qualitative match has been found between them. These
event-shape differentiated symmetry plane correlations can be used to deduce
the presence of higher-order anisotropies in the initial energy distribution,
thus giving insight into the initial geometry of the colliding system, among
other applications like model development and model testing using Bayesian
analyses.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:43:58 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Tripathy",
"Sarthak",
""
],
[
"Prasad",
"Suraj",
""
],
[
"Sahoo",
"Raghunath",
""
]
] |
2504.09276 | Xiyue Han | Xiyue Han and Alexander Schied | On the rate of convergence of estimating the Hurst parameter of rough
stochastic volatility models | 10 pages | null | null | null | q-fin.ST math.PR math.ST q-fin.MF stat.TH | http://creativecommons.org/licenses/by/4.0/ | In [8], easily computable scale-invariant estimator
$\widehat{\mathscr{R}}^s_n$ was constructed to estimate the Hurst parameter of
the drifted fractional Brownian motion $X$ from its antiderivative. This paper
extends this convergence result by proving that $\widehat{\mathscr{R}}^s_n$
also consistently estimates the Hurst parameter when applied to the
antiderivative of $g \circ X$ for a general nonlinear function $g$. We also
establish an almost sure rate of convergence in this general setting. Our
result applies, in particular, to the estimation of the Hurst parameter of a
wide class of rough stochastic volatility models from discrete observations of
the integrated variance, including the fractional stochastic volatility model.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:48:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Han",
"Xiyue",
""
],
[
"Schied",
"Alexander",
""
]
] |
2504.09277 | Ashmi Banerjee | Ashmi Banerjee, Adithi Satish, Fitri Nur Aisyah, Wolfgang W\"orndl,
and Yashar Deldjoo | SynthTRIPs: A Knowledge-Grounded Framework for Benchmark Query
Generation for Personalized Tourism Recommenders | Accepted for publication at SIGIR 2025 | null | 10.1145/3726302.3730321 | null | cs.IR cs.AI | http://creativecommons.org/licenses/by/4.0/ | Tourism Recommender Systems (TRS) are crucial in personalizing travel
experiences by tailoring recommendations to users' preferences, constraints,
and contextual factors. However, publicly available travel datasets often lack
sufficient breadth and depth, limiting their ability to support advanced
personalization strategies -- particularly for sustainable travel and off-peak
tourism. In this work, we explore using Large Language Models (LLMs) to
generate synthetic travel queries that emulate diverse user personas and
incorporate structured filters such as budget constraints and sustainability
preferences.
This paper introduces a novel SynthTRIPs framework for generating synthetic
travel queries using LLMs grounded in a curated knowledge base (KB). Our
approach combines persona-based preferences (e.g., budget, travel style) with
explicit sustainability filters (e.g., walkability, air quality) to produce
realistic and diverse queries. We mitigate hallucination and ensure factual
correctness by grounding the LLM responses in the KB. We formalize the query
generation process and introduce evaluation metrics for assessing realism and
alignment. Both human expert evaluations and automatic LLM-based assessments
demonstrate the effectiveness of our synthetic dataset in capturing complex
personalization aspects underrepresented in existing datasets. While our
framework was developed and tested for personalized city trip recommendations,
the methodology applies to other recommender system domains.
Code and dataset are made public at https://bit.ly/synthTRIPs
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:48:35 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Banerjee",
"Ashmi",
""
],
[
"Satish",
"Adithi",
""
],
[
"Aisyah",
"Fitri Nur",
""
],
[
"Wörndl",
"Wolfgang",
""
],
[
"Deldjoo",
"Yashar",
""
]
] |
2504.09278 | Nitesh Kumar | Nitesh Kumar (1), Philippe Prugniel (2), and Harinder P. Singh (3)
((1) Department of Physics, Cluster of Applied Science, University of
Petroleum and Energy Studies (UPES), Bidholi, Dehradun, 248007, Uttarakhand,
(2) Universit\'e de Lyon, Universit\'e Lyon 1, 69622 Villeurbanne, CRAL,
Observatoire de Lyon, CNRS UMR 5574, 69561 Saint-Genis Laval, France, (3)
Department of Physics and Astrophysics, University of Delhi, Delhi 110007,
India.) | Physical Parameters of Stars in NGC 6397 Using ANN-Based Interpolation
and Full Spectrum Fitting | Accepted for Publication in Journal New Astronomy | null | null | null | astro-ph.SR astro-ph.GA astro-ph.IM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Stellar spectral interpolation is critical technique employed by fitting
software to derive the physical parameters of stars. This approach is necessary
because on-the-go generation of synthetic stellar spectra is not possible due
to the complex and high cost of computation. The goal of this study is to
develop a spectral interpolator for a synthetic spectral library using
artificial neural networks (ANNs). The study aims to test the accuracy of the
trained interpolator through self-inversion and, subsequently, to utilize the
interpolator to derive the physical parameters of stars in the globular cluster
NGC 6397 using spectra obtained from the Multi Unit Spectroscopic Explorer
(MUSE) on the Very Large Telescope (VLT). In this study, ANNs were trained to
function as spectral interpolators. The ULySS full-spectrum fitting package,
integrated with the trained interpolators, was then used to extract the
physical parameters of 1587 spectra of 1063 stars in NGC 6397. The trained ANN
interpolator achieved precise determination of stellar parameters with a mean
difference of 31 K for $T_{\rm eff}$ and 0.01 dex for [Fe/H] compared to
previous studies. This study demonstrates the efficacy of ANN-based spectral
interpolation in stellar parameter determination, offering faster and more
accurate analysis.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:50:09 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Kumar",
"Nitesh",
""
],
[
"Prugniel",
"Philippe",
""
],
[
"Singh",
"Harinder P.",
""
]
] |
2504.09279 | Tengyuan Liang | Nabarun Deb, Tengyuan Liang | No-Regret Generative Modeling via Parabolic Monge-Amp\`ere PDE | 30 pages, 3 figures | null | null | null | stat.ML cs.LG math.OC math.ST stat.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a novel generative modeling framework based on a discretized
parabolic Monge-Amp\`ere PDE, which emerges as a continuous limit of the
Sinkhorn algorithm commonly used in optimal transport. Our method performs
iterative refinement in the space of Brenier maps using a mirror gradient
descent step. We establish theoretical guarantees for generative modeling
through the lens of no-regret analysis, demonstrating that the iterates
converge to the optimal Brenier map under a variety of step-size schedules. As
a technical contribution, we derive a new Evolution Variational Inequality
tailored to the parabolic Monge-Amp\`ere PDE, connecting geometry,
transportation cost, and regret. Our framework accommodates non-log-concave
target distributions, constructs an optimal sampling process via the Brenier
map, and integrates favorable learning techniques from generative adversarial
networks and score-based diffusion models. As direct applications, we
illustrate how our theory paves new pathways for generative modeling and
variational inference.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:51:02 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Deb",
"Nabarun",
""
],
[
"Liang",
"Tengyuan",
""
]
] |
2504.09280 | Peng-Cheng Hang | Peng-Cheng Hang, Liangjian Hu | Full asymptotic expansions of the Humbert function $\Phi_1$ | 12 pages | null | null | null | math.CA math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We derive full asymptotic expansions for the Humbert function $\Phi_1$ in
different limiting regimes of its variables. Our derivation employs various
asymptotic methods and relies on key transformation formulae established by
Erd\'elyi (1940), and Tuan and Kalla (1987). The efficiency of our asymptotic
results are also illustrated through two applications: (1) analytic
continuations of Saran's function $F_M$, and (2) two limits arising in the
study of the $1D$ Glauber-Ising model. Finally, some promising directions for
future research are highlighted.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 16:51:31 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hang",
"Peng-Cheng",
""
],
[
"Hu",
"Liangjian",
""
]
] |
2504.09282 | Zheyuan Zhang | Zheyuan Zhang, Monica Dou, Linkai Peng, Hongyi Pan, Ulas Bagci, Boqing
Gong | VideoAds for Fast-Paced Video Understanding: Where Opensource Foundation
Models Beat GPT-4o & Gemini-1.5 Pro | null | null | null | null | cs.CV | http://creativecommons.org/publicdomain/zero/1.0/ | Advertisement videos serve as a rich and valuable source of purpose-driven
information, encompassing high-quality visual, textual, and contextual cues
designed to engage viewers. They are often more complex than general videos of
similar duration due to their structured narratives and rapid scene
transitions, posing significant challenges to multi-modal large language models
(MLLMs). In this work, we introduce VideoAds, the first dataset tailored for
benchmarking the performance of MLLMs on advertisement videos. VideoAds
comprises well-curated advertisement videos with complex temporal structures,
accompanied by \textbf{manually} annotated diverse questions across three core
tasks: visual finding, video summary, and visual reasoning. We propose a
quantitative measure to compare VideoAds against existing benchmarks in terms
of video complexity. Through extensive experiments, we find that
Qwen2.5-VL-72B, an opensource MLLM, achieves 73.35\% accuracy on VideoAds,
outperforming GPT-4o (66.82\%) and Gemini-1.5 Pro (69.66\%); the two
proprietary models especially fall behind the opensource model in video
summarization and reasoning, but perform the best in visual finding. Notably,
human experts easily achieve a remarkable accuracy of 94.27\%. These results
underscore the necessity of advancing MLLMs' temporal modeling capabilities and
highlight VideoAds as a potentially pivotal benchmark for future research in
understanding video that requires high FPS sampling. The dataset and evaluation
code will be publicly available at https://videoadsbenchmark.netlify.app.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:05:35 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhang",
"Zheyuan",
""
],
[
"Dou",
"Monica",
""
],
[
"Peng",
"Linkai",
""
],
[
"Pan",
"Hongyi",
""
],
[
"Bagci",
"Ulas",
""
],
[
"Gong",
"Boqing",
""
]
] |
2504.09283 | Ian Arawjo | Priyan Vaithilingam, Munyeong Kim, Frida-Cecilia Acosta-Parenteau,
Daniel Lee, Amine Mhedhbi, Elena L. Glassman, Ian Arawjo | Semantic Commit: Helping Users Update Intent Specifications for AI
Memory at Scale | 22 pages; 10 figures | null | null | null | cs.HC cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | How do we update AI memory of user intent as intent changes? We consider how
an AI interface may assist the integration of new information into a repository
of natural language data. Inspired by software engineering concepts like impact
analysis, we develop methods and a UI for managing semantic changes with
non-local effects, which we call "semantic conflict resolution." The user
commits new intent to a project -- makes a "semantic commit" -- and the AI
helps the user detect and resolve semantic conflicts within a store of existing
information representing their intent (an "intent specification"). We develop
an interface, SemanticCommit, to better understand how users resolve conflicts
when updating intent specifications such as Cursor Rules and game design
documents. A knowledge graph-based RAG pipeline drives conflict detection,
while LLMs assist in suggesting resolutions. We evaluate our technique on an
initial benchmark. Then, we report a 12 user within-subjects study of
SemanticCommit for two task domains -- game design documents, and AI agent
memory in the style of ChatGPT memories -- where users integrated new
information into an existing list. Half of our participants adopted a workflow
of impact analysis, where they would first flag conflicts without AI revisions
then resolve conflicts locally, despite having access to a global revision
feature. We argue that AI agent interfaces, such as software IDEs like Cursor
and Windsurf, should provide affordances for impact analysis and help users
validate AI retrieval independently from generation. Our work speaks to how AI
agent designers should think about updating memory as a process that involves
human feedback and decision-making.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:07:08 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Vaithilingam",
"Priyan",
""
],
[
"Kim",
"Munyeong",
""
],
[
"Acosta-Parenteau",
"Frida-Cecilia",
""
],
[
"Lee",
"Daniel",
""
],
[
"Mhedhbi",
"Amine",
""
],
[
"Glassman",
"Elena L.",
""
],
[
"Arawjo",
"Ian",
""
]
] |
2504.09284 | Zuyi Zhang | Zuyi Zhang | Uniqueness of holomorphic quilts lifted from holomorphic bigons on
surfaces | 18 pages 6 figures | null | null | null | math.SG math.GT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the author's previous paper, the author constructed holomorphic quilts
from the bigons of the Lagrangian Floer chain group after performing Lagrangian
composition. This paper proves the uniqueness of such holomorphic quilts. As a
consequence, it provides a combinatorial method for computing the boundary map
of immersed Lagrangian Floer chain groups when the symplectic manifolds are
closed surfaces. One outcome is the construction of many examples exhibiting
figure eight bubbling, which also confirms a conjecture of Cazassus Herald Kirk
Kotelskiy.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:07:54 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhang",
"Zuyi",
""
]
] |
2504.09285 | Chaoyi Ruan | Chaoyi Ruan, Yinhe Chen, Dongqi Tian, Yandong Shi, Yongji Wu, Jialin
Li, Cheng Li | DynaServe: Unified and Elastic Tandem-Style Execution for Dynamic
Disaggregated LLM Serving | null | null | null | null | cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Modern large language model (LLM) serving must efficiently handle highly
dynamic workloads, where prompt and response lengths vary significantly across
requests. Existing systems typically adopt either colocated execution, where
prefill and decode stages share the same GPU for high throughput, or
disaggregated execution, which decouples the two stages and assign their tasks
to dedicated GPUs for interference avoidance. However, both paradigms face
critical limitations: colocation suffers from resource contention and prolonged
tail latency, whereas disaggregation likely leads to resource wasting when
prefill or decode GPUs are not fully occupied.
To address the above limitations, we introduce DynaServe, a unified LLM
serving framework based on the Tandem Serving model. Under this model,
DynaServe elastically decomposes each request into two virtual sub-requests
that are collaboratively processed by a pair of GPU instances. The Lead GPU
handles the initial prompt and early generation, while the Follow GPU completes
decoding, enabling dynamic load balancing, fine-grained batching, and coherent
execution across distributed resources. By coordinating computation and memory
across the cluster, DynaServe adapts to diverse and bursty workloads while
maintaining stringent latency service-level objectives (SLOs). Evaluations on
real-world traces show that DynaServe improves end-to-end Serving Capacity by
up to 1.23 $\times$, increases the overall goodput from 1.15 $\times$ to 4.34
$\times$, and improve the memory utilization by up to 49% compared to
state-of-the-art colocated and disaggregated systems.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:09:54 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ruan",
"Chaoyi",
""
],
[
"Chen",
"Yinhe",
""
],
[
"Tian",
"Dongqi",
""
],
[
"Shi",
"Yandong",
""
],
[
"Wu",
"Yongji",
""
],
[
"Li",
"Jialin",
""
],
[
"Li",
"Cheng",
""
]
] |
2504.09286 | Florian Eisele | Florian Eisele, Ricardo J. Franquiz Flores, John W. MacQuarrie | Block pro-fusion systems for profinite groups and blocks with infinite
dihedral defect groups | Comments welcome | null | null | null | math.RT math.GR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce block pro-fusion systems for blocks of profinite groups, prove a
profinite version of Puig's structure theorem for nilpotent blocks, and use it
to show that there is only one Morita equivalence class of blocks having the
infinite dihedral pro-$2$ group as their defect group.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:10:10 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Eisele",
"Florian",
""
],
[
"Flores",
"Ricardo J. Franquiz",
""
],
[
"MacQuarrie",
"John W.",
""
]
] |
2504.09287 | Yuri Kovalev | Y. Y. Kovalev (MPIfR), A. B. Pushkarev (CrAO, Lebedev, INR), J. L.
Gomez (IAA-CSIC), D. C. Homan (Denison U), M. L. Lister (Purdue U), J. D.
Livingston (MPIfR), I. N. Pashchenko (Lebedev), A. V. Plavin (Harvard U), T.
Savolainen (Aalto U, MPIfR), S. V. Troitsky (INR, MSU) | Looking into the Jet Cone of the Neutrino-Associated Very High Energy
Blazar PKS 1424+240 | 7 pages, 5 figures | null | null | INR-TH-2025-004 | astro-ph.HE astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The acceleration process of massive particles as well as the production of
very high energy (VHE) photons and neutrinos remains a fundamental challenge in
astrophysics. We investigate the parsec-scale properties of the blazar PKS
1424+240, that has been selected on the basis of both gamma-ray and neutrino
VHE radiation. We analyze VLBA observations of this BL Lac object, stacking 42
polarization-sensitive images collected in 2009-2025 to enhance the signal and
reveal persistent parsec-scale structure. Our observations indicate that this
object is viewed inside the jet cone, very close to the axis of its
relativistic jet, with a viewing angle of <0.6 deg. This effectively maximizes
Doppler boosting to values ~30 and enhances both electromagnetic and neutrino
emission in the observer's direction. Based on polarimetric observations, we
unambiguously detect a net toroidal component of the jet's magnetic field.
Blazars with very small jet viewing angles offer a solution to the longstanding
mismatch between Doppler factors inferred from VLBI and those derived from VHE
observations -- the so-called `Doppler factor crisis'. We show that
relativistic beaming plays the critical role in the gamma-ray and neutrino
emission of blazars, with direct implications for models of their
multi-messenger emission.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:14:47 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Kovalev",
"Y. Y.",
"",
"MPIfR"
],
[
"Pushkarev",
"A. B.",
"",
"CrAO, Lebedev, INR"
],
[
"Gomez",
"J. L.",
"",
"IAA-CSIC"
],
[
"Homan",
"D. C.",
"",
"Denison U"
],
[
"Lister",
"M. L.",
"",
"Purdue U"
],
[
"Livingston",
"J. D.",
"",
"MPIfR"
],
[
"Pashchenko",
"I. N.",
"",
"Lebedev"
],
[
"Plavin",
"A. V.",
"",
"Harvard U"
],
[
"Savolainen",
"T.",
"",
"Aalto U, MPIfR"
],
[
"Troitsky",
"S. V.",
"",
"INR, MSU"
]
] |
2504.09288 | Kushagra Parashar | Kushagra Parashar, Ajay Dev, Aditya Kumar, Darpan Khatri | Enhancing Productivity in Database Management Through AI: A Three-Phase
Approach for Database | 7 pages, 1 figure. Submitted for academic review at ABV-IIITM
Gwalior. Includes AI integration with PostgreSQL, real-world deployment case
studies, and productivity benchmarks | null | null | null | cs.DB | http://creativecommons.org/licenses/by/4.0/ | This paper presents a novel AI-powered framework designed to streamline
database management and query optimization for PostgreSQL systems. Structured
in three phases: Natural Language to SQL Translation, Query Execution and
Analysis, and Insight Generation, the approach empowers users with intuitive,
intelligent interaction with databases. By leveraging advanced natural language
processing and language models, the system enables non-technical users to
extract meaningful insights from complex datasets, reducing the dependency on
specialized DBAs. The framework also introduces iterative refinement of
queries, automatic report generation, and support for temporal data
forecasting. Experimental results demonstrate improved productivity, reduced
query latency, and enhanced accuracy, validating the system's effectiveness
across diverse business use cases. The solution was developed and evaluated at
ABV-IIITM Gwalior, under the guidance of Prof. Dr. Arun Kuma
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:18:28 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Parashar",
"Kushagra",
""
],
[
"Dev",
"Ajay",
""
],
[
"Kumar",
"Aditya",
""
],
[
"Khatri",
"Darpan",
""
]
] |
2504.09289 | Konstantinos Fotopoulos | Konstantinos Fotopoulos, Christos Garoufis, Petros Maragos | Sparse Hybrid Linear-Morphological Networks | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate hybrid linear-morphological networks. Recent studies highlight
the inherent affinity of morphological layers to pruning, but also their
difficulty in training. We propose a hybrid network structure, wherein
morphological layers are inserted between the linear layers of the network, in
place of activation functions. We experiment with the following morphological
layers: 1) maxout pooling layers (as a special case of a morphological layer),
2) fully connected dense morphological layers, and 3) a novel, sparsely
initialized variant of (2). We conduct experiments on the Magna-Tag-A-Tune
(music auto-tagging) and CIFAR-10 (image classification) datasets, replacing
the linear classification heads of state-of-the-art convolutional network
architectures with our proposed network structure for the various morphological
layers. We demonstrate that these networks induce sparsity to their linear
layers, making them more prunable under L1 unstructured pruning. We also show
that on MTAT our proposed sparsely initialized layer achieves slightly better
performance than ReLU, maxout, and densely initialized max-plus layers, and
exhibits faster initial convergence.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:19:46 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Fotopoulos",
"Konstantinos",
""
],
[
"Garoufis",
"Christos",
""
],
[
"Maragos",
"Petros",
""
]
] |
2504.09290 | Alexandr Dodin | M. A. Burlak, K. N. Grankin, A. V. Dodin, A. V. Zharova, N. P.
Ikonnikova, V. A. Kiryukhina, S. A. Lamzin, B. S. Safonov, I. A. Strakhov | On the causes of brightness variability of the young star BP Tau | 15 pages, 10 figures, Accepted by Astrophysical Bulletin | null | null | null | astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | We have constructed and analysed the secular light curve of BP Tau, a
classical T Tauri-type star. Wave-like variations in the average brightness
were detected, with an amplitude of $\Delta B\approx 0.2$ and characteristic
time-scales of several decades. We argue that three deep dimming events
$(\Delta B \sim 1.5)$, lasting from 1 hour to several days, are caused by the
eclipse of a hot (accretion) spot by dust falling onto the star together with
gas. Such eclipses, albeit with smaller amplitudes, may explain the absence of
a strictly defined periodicity in the brightness variations of BP Tau
associated with axial rotation. We also show that within the distance range of
0.1 to 200 AU, BP Tau does not have a companion with a mass exceeding
0.2M$_\odot.$ The causes of brightness and colour index variations on different
time-scales are discussed.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:21:47 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Burlak",
"M. A.",
""
],
[
"Grankin",
"K. N.",
""
],
[
"Dodin",
"A. V.",
""
],
[
"Zharova",
"A. V.",
""
],
[
"Ikonnikova",
"N. P.",
""
],
[
"Kiryukhina",
"V. A.",
""
],
[
"Lamzin",
"S. A.",
""
],
[
"Safonov",
"B. S.",
""
],
[
"Strakhov",
"I. A.",
""
]
] |
2504.09291 | Jiaying Qian | Jiaying Qian, Ziheng Jia, Zicheng Zhang, Zeyu Zhang, Guangtao Zhai,
Xiongkuo Min | Towards Explainable Partial-AIGC Image Quality Assessment | null | null | null | null | cs.CV cs.MM | http://creativecommons.org/licenses/by/4.0/ | The rapid advancement of AI-driven visual generation technologies has
catalyzed significant breakthroughs in image manipulation, particularly in
achieving photorealistic localized editing effects on natural scene images
(NSIs). Despite extensive research on image quality assessment (IQA) for
AI-generated images (AGIs), most studies focus on fully AI-generated outputs
(e.g., text-to-image generation), leaving the quality assessment of
partial-AIGC images (PAIs)-images with localized AI-driven edits an almost
unprecedented field. Motivated by this gap, we construct the first large-scale
PAI dataset towards explainable partial-AIGC image quality assessment (EPAIQA),
the EPAIQA-15K, which includes 15K images with localized AI manipulation in
different regions and over 300K multi-dimensional human ratings. Based on this,
we leverage large multi-modal models (LMMs) and propose a three-stage model
training paradigm. This paradigm progressively trains the LMM for editing
region grounding, quantitative quality scoring, and quality explanation.
Finally, we develop the EPAIQA series models, which possess explainable quality
feedback capabilities. Our work represents a pioneering effort in the
perceptual IQA field for comprehensive PAI quality assessment.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:27:50 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Qian",
"Jiaying",
""
],
[
"Jia",
"Ziheng",
""
],
[
"Zhang",
"Zicheng",
""
],
[
"Zhang",
"Zeyu",
""
],
[
"Zhai",
"Guangtao",
""
],
[
"Min",
"Xiongkuo",
""
]
] |
2504.09292 | Rajesh Selukar | Rajesh Selukar | Stochastic Claims Reserving Using State Space Modeling | null | null | null | null | stat.CO | http://creativecommons.org/licenses/by/4.0/ | Claims reserving, also known as Incurred But Not Reported (IBNR) claims
prediction, is an important issue in general insurance. State space modeling is
widely recognized as a statistically robust method for addressing this problem.
In state space model-based claims reserving, the Kalman filter and Kalman
smoother algorithms are employed for model fitting, diagnostics, and deriving
reserve estimates. Additionally, the simulation smoother algorithm is used to
obtain the sampling distribution of the derived reserve estimate. The
integration of these three algorithms results in an elegant and transparent
claim reserving process. Various state space models (SSMs) have been proposed
in the literature for claims reserving. This article outlines a step-by-step
process for computing the SSM-based reserve estimate and its associated
sampling distribution for any proposed SSM. A brief discussion on model
selection is also included. The claims reserving computations are demonstrated
using a real-life data set. The state space modeling computations in the
illustrations are performed by using the CSSM procedure in SAS
Viya/Econometrics software. The SAS code for reproducing the output in the
illustrations is provided in the supplementary material.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:29:29 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Selukar",
"Rajesh",
""
]
] |
2504.09293 | Daniel \'Alvarez | Daniel \'Alvarez | Morita equivalences, moduli spaces and flag varieties | null | null | null | null | math.SG math.AG math.DG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Double Bruhat cells in a complex semisimple Lie group $G$ emerged as a
crucial concept in the work of S. Fomin and A. Zelevinsky on total positivity
and cluster algebras. Double Bruhat cells are special instances of a broader
class of cluster varieties known as generalized double Bruhat cells. These can
be studied collectively as Poisson subvarieties of $\widetilde{F}_{2n} = G
\times \mathcal{B}^{2n-1}$, where $\mathcal{B}$ is the flag variety of $G$. The
spaces $\widetilde{F}_{2n}$ are Poisson groupoids over $\mathcal{B}^n$, and
they were introduced in the study of configuration Poisson groupoids of flags
by J.-H. Lu, V. Mouquin, and S. Yu.
In this work, we describe the spaces $\widetilde{F}_{2n}$ as decorated moduli
spaces of flat $G$-bundles over a disc. As a consequence, we obtain the
following results. (1) We explicitly integrate the Poisson groupoids
$\widetilde{F}_{2n}$ to double symplectic groupoids, which are complex
algebraic varieties. Moreover, we show that these integrations are
symplectically Morita equivalent for all $n$, thereby recovering the Poisson
bimodule structures on double Bruhat cells via restriction. (2) Using the
previous construction, we integrate the Poisson subgroupoids of
$\widetilde{F}_{2n}$ given by unions of generalized double Bruhat cells to
explicit double symplectic groupoids. As a corollary, we obtain integrations of
the top-dimensional generalized double Bruhat cells inside them. (3) Finally,
we relate our integration with the work of P. Boalch on meromorphic
connections. We lift to the groupoids the torus actions that give rise to such
cluster varieties and show that they correspond to the quasi-Hamiltonian
actions on the fission spaces of irregular singularities.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:29:40 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Álvarez",
"Daniel",
""
]
] |
2504.09294 | Hanyu Jin | Hanyu Jin, Zhefan Xu, Haoyu Shen, Xinming Han, Kanlong Ye and Kenji
Shimada | Adaptive Planning Framework for UAV-Based Surface Inspection in
Partially Unknown Indoor Environments | null | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Inspecting indoor environments such as tunnels, industrial facilities, and
construction sites is essential for infrastructure monitoring and maintenance.
While manual inspection in these environments is often time-consuming and
potentially hazardous, Unmanned Aerial Vehicles (UAVs) can improve efficiency
by autonomously handling inspection tasks. Such inspection tasks usually rely
on reference maps for coverage planning. However, in industrial applications,
only the floor plans are typically available. The unforeseen obstacles not
included in the floor plans will result in outdated reference maps and
inefficient or unsafe inspection trajectories. In this work, we propose an
adaptive inspection framework that integrates global coverage planning with
local reactive adaptation to improve the coverage and efficiency of UAV-based
inspection in partially unknown indoor environments. Experimental results in
structured indoor scenarios demonstrate the effectiveness of the proposed
approach in inspection efficiency and achieving high coverage rates with
adaptive obstacle handling, highlighting its potential for enhancing the
efficiency of indoor facility inspection.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:30:11 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Jin",
"Hanyu",
""
],
[
"Xu",
"Zhefan",
""
],
[
"Shen",
"Haoyu",
""
],
[
"Han",
"Xinming",
""
],
[
"Ye",
"Kanlong",
""
],
[
"Shimada",
"Kenji",
""
]
] |
2504.09295 | Raon\'i Ponciano | Jo\~ao Marcos do \'O, Jos\'e Francisco de Oliveira, and Raon\'i Cabral
Ponciano | Trudinger-Moser type inequalities for the Hessian equation with
logarithmic weights | null | null | null | null | math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We establish sharp Trudinger-Moser inequalities with logarithmic weights for
the $k$-Hessian equation and investigate the existence of maximizers. Our
analysis extends the classical results of Tian and Wang to $k$-admissible
function spaces with logarithmic weights, providing a natural complement to the
work of Calanchi and Ruf. Our approach relies on transforming the problem into
a one-dimensional weighted Sobolev space, where we solve it using various
techniques, including some radial lemmas and certain Hardy-type inequalities,
which we establish in this paper, as well as a theorem due to Leckband.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:35:46 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ó",
"João Marcos do",
""
],
[
"de Oliveira",
"José Francisco",
""
],
[
"Ponciano",
"Raoní Cabral",
""
]
] |
2504.09296 | Qing Zhang | Zhang Qing, Rekimoto Jun | Look and Talk: Seamless AI Assistant Interaction with Gaze-Triggered
Activation | null | null | null | null | cs.HC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Engaging with AI assistants to gather essential information in time is
becoming increasingly common. Traditional activation methods, like wake words
such as Hey Siri, Ok Google, and Hey Alexa, are constrained by technical
challenges such as false activations, recognition errors, and discomfort in
public settings. Similarly, activating AI systems via physical buttons imposes
strict interactive limitations as it demands particular physical actions, which
hinders fluid and spontaneous communication with AI. Our approach employs
eye-tracking technology within AR glasses to discern a user's intention to
engage with the AI assistant. By sustaining eye contact on a virtual AI avatar
for a specific time, users can initiate an interaction silently and without
using their hands. Preliminary user feedback suggests that this technique is
relatively intuitive, natural, and less obtrusive, highlighting its potential
for integrating AI assistants fluidly into everyday interactions.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:39:19 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Qing",
"Zhang",
""
],
[
"Jun",
"Rekimoto",
""
]
] |
2504.09297 | Huu-Phong Phan-Nguyen | Huu-Phong Phan-Nguyen, Anh Dao, Tien-Huy Nguyen, Tuan Quang, Huu-Loc
Tran, Tinh-Anh Nguyen-Nhu, Huy-Thach Pham, Quan Nguyen, Hoang M. Le,
Quang-Vinh Dinh | Cycle Training with Semi-Supervised Domain Adaptation: Bridging Accuracy
and Efficiency for Real-Time Mobile Scene Detection | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Nowadays, smartphones are ubiquitous, and almost everyone owns one. At the
same time, the rapid development of AI has spurred extensive research on
applying deep learning techniques to image classification. However, due to the
limited resources available on mobile devices, significant challenges remain in
balancing accuracy with computational efficiency. In this paper, we propose a
novel training framework called Cycle Training, which adopts a three-stage
training process that alternates between exploration and stabilization phases
to optimize model performance. Additionally, we incorporate Semi-Supervised
Domain Adaptation (SSDA) to leverage the power of large models and unlabeled
data, thereby effectively expanding the training dataset. Comprehensive
experiments on the CamSSD dataset for mobile scene detection demonstrate that
our framework not only significantly improves classification accuracy but also
ensures real-time inference efficiency. Specifically, our method achieves a
94.00% in Top-1 accuracy and a 99.17% in Top-3 accuracy and runs inference in
just 1.61ms using CPU, demonstrating its suitability for real-world mobile
deployment.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:42:45 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Phan-Nguyen",
"Huu-Phong",
""
],
[
"Dao",
"Anh",
""
],
[
"Nguyen",
"Tien-Huy",
""
],
[
"Quang",
"Tuan",
""
],
[
"Tran",
"Huu-Loc",
""
],
[
"Nguyen-Nhu",
"Tinh-Anh",
""
],
[
"Pham",
"Huy-Thach",
""
],
[
"Nguyen",
"Quan",
""
],
[
"Le",
"Hoang M.",
""
],
[
"Dinh",
"Quang-Vinh",
""
]
] |
2504.09298 | Huu-Phong Phan-Nguyen | Tinh-Anh Nguyen-Nhu, Huu-Loc Tran, Nguyen-Khang Le, Minh-Nhat Nguyen,
Tien-Huy Nguyen, Hoang-Long Nguyen-Huu, Huu-Phong Phan-Nguyen, Huy-Thach
Pham, Quan Nguyen, Hoang M. Le, Quang-Vinh Dinh | A Lightweight Moment Retrieval System with Global Re-Ranking and Robust
Adaptive Bidirectional Temporal Search | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | The exponential growth of digital video content has posed critical challenges
in moment-level video retrieval, where existing methodologies struggle to
efficiently localize specific segments within an expansive video corpus.
Current retrieval systems are constrained by computational inefficiencies,
temporal context limitations, and the intrinsic complexity of navigating video
content. In this paper, we address these limitations through a novel
Interactive Video Corpus Moment Retrieval framework that integrates a
SuperGlobal Reranking mechanism and Adaptive Bidirectional Temporal Search
(ABTS), strategically optimizing query similarity, temporal stability, and
computational resources. By preprocessing a large corpus of videos using a
keyframe extraction model and deduplication technique through image hashing,
our approach provides a scalable solution that significantly reduces storage
requirements while maintaining high localization precision across diverse video
repositories.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 17:49:46 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Nguyen-Nhu",
"Tinh-Anh",
""
],
[
"Tran",
"Huu-Loc",
""
],
[
"Le",
"Nguyen-Khang",
""
],
[
"Nguyen",
"Minh-Nhat",
""
],
[
"Nguyen",
"Tien-Huy",
""
],
[
"Nguyen-Huu",
"Hoang-Long",
""
],
[
"Phan-Nguyen",
"Huu-Phong",
""
],
[
"Pham",
"Huy-Thach",
""
],
[
"Nguyen",
"Quan",
""
],
[
"Le",
"Hoang M.",
""
],
[
"Dinh",
"Quang-Vinh",
""
]
] |
2504.09299 | Sonia Laguna | Marco Voegeli, Sonia Laguna, Heike Leutheuser, Marc Pfister,
Marie-Anne Burckhardt, Julia E Vogt | Beyond Glucose-Only Assessment: Advancing Nocturnal Hypoglycemia
Prediction in Children with Type 1 Diabetes | Published at ICLR 2025 Workshop on AI for Children | null | null | null | cs.LG q-bio.QM | http://creativecommons.org/licenses/by/4.0/ | The dead-in-bed syndrome describes the sudden and unexplained death of young
individuals with Type 1 Diabetes (T1D) without prior long-term complications.
One leading hypothesis attributes this phenomenon to nocturnal hypoglycemia
(NH), a dangerous drop in blood glucose during sleep. This study aims to
improve NH prediction in children with T1D by leveraging physiological data and
machine learning (ML) techniques. We analyze an in-house dataset collected from
16 children with T1D, integrating physiological metrics from wearable sensors.
We explore predictive performance through feature engineering, model selection,
architectures, and oversampling. To address data limitations, we apply transfer
learning from a publicly available adult dataset. Our results achieve an AUROC
of 0.75 +- 0.21 on the in-house dataset, further improving to 0.78 +- 0.05 with
transfer learning. This research moves beyond glucose-only predictions by
incorporating physiological parameters, showcasing the potential of ML to
enhance NH detection and improve clinical decision-making for pediatric
diabetes management.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:07:40 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Voegeli",
"Marco",
""
],
[
"Laguna",
"Sonia",
""
],
[
"Leutheuser",
"Heike",
""
],
[
"Pfister",
"Marc",
""
],
[
"Burckhardt",
"Marie-Anne",
""
],
[
"Vogt",
"Julia E",
""
]
] |
2504.09300 | Jesse Selover | Jeffrey Chen, Jesse Selover | Positivity properties of $q$-hit numbers in the finite general linear
group | 19 pages, 2 figures | null | null | null | math.CO | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We consider the problem of counting matrices over a finite field with fixed
rank and support contained in a fixed set. The count of such matrices gives a
$q$-analogue of the classical rook and hit numbers, known as the $q$-rook and
$q$-hit numbers. They are known not to be polynomial in $q$ in general. We use
inclusion-exclusion on the support of the matrices and the orbit counting
method of Lewis et al. to show that the residues of these functions in low
degrees are polynomial. We define a generalization of the classical rook and
hit numbers which count placements of certain classes of graphs. These give us
a formula for residues of the $q$-rook and $q$-hit numbers in low degrees. We
analyze the residues of the $q$-hit number and show that the coefficient of
$q-1$ in the $q$-hit number is always non-negative.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:09:02 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Chen",
"Jeffrey",
""
],
[
"Selover",
"Jesse",
""
]
] |
2504.09301 | Pengcheng Zhou | Pengcheng Zhou, Zhiqiang Nie, Haochen Li | Continuum-Interaction-Driven Intelligence: Human-Aligned Neural
Architecture via Crystallized Reasoning and Fluid Generation | null | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Current AI systems based on probabilistic neural networks, such as large
language models (LLMs), have demonstrated remarkable generative capabilities
yet face critical challenges including hallucination, unpredictability, and
misalignment with human decision-making. These issues fundamentally stem from
the over-reliance on randomized (probabilistic) neural networks-oversimplified
models of biological neural networks-while neglecting the role of procedural
reasoning (chain-of-thought) in trustworthy decision-making. Inspired by the
human cognitive duality of fluid intelligence (flexible generation) and
crystallized intelligence (structured knowledge), this study proposes a
dual-channel intelligent architecture that integrates probabilistic generation
(LLMs) with white-box procedural reasoning (chain-of-thought) to construct
interpretable, continuously learnable, and human-aligned AI systems.
Concretely, this work: (1) redefines chain-of-thought as a programmable
crystallized intelligence carrier, enabling dynamic knowledge evolution and
decision verification through multi-turn interaction frameworks; (2) introduces
a task-driven modular network design that explicitly demarcates the functional
boundaries between randomized generation and procedural control to address
trustworthiness in vertical-domain applications; (3) demonstrates that
multi-turn interaction is a necessary condition for intelligence emergence,
with dialogue depth positively correlating with the system's human-alignment
degree. This research not only establishes a new paradigm for trustworthy AI
deployment but also provides theoretical foundations for next-generation
human-AI collaborative systems.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:15:49 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhou",
"Pengcheng",
""
],
[
"Nie",
"Zhiqiang",
""
],
[
"Li",
"Haochen",
""
]
] |
2504.09302 | Junichiro Takahashi | Junichiro Takahashi, JingChuan Guan, Masataka Sato, Kaito Baba, Kazuto
Haruguchi, Daichi Nagashima, Satoshi Kodera, Norihiko Takeda | Application of Contrastive Learning on ECG Data: Evaluating Performance
in Japanese and Classification with Around 100 Labels | 13 pages, 1 figures | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | The electrocardiogram (ECG) is a fundamental tool in cardiovascular
diagnostics due to its powerful and non-invasive nature. One of the most
critical usages is to determine whether more detailed examinations are
necessary, with users ranging across various levels of expertise. Given this
diversity in expertise, it is essential to assist users to avoid critical
errors. Recent studies in machine learning have addressed this challenge by
extracting valuable information from ECG data. Utilizing language models, these
studies have implemented multimodal models aimed at classifying ECGs according
to labeled terms. However, the number of classes was reduced, and it remains
uncertain whether the technique is effective for languages other than English.
To move towards practical application, we utilized ECG data from regular
patients visiting hospitals in Japan, maintaining a large number of Japanese
labels obtained from actual ECG readings. Using a contrastive learning
framework, we found that even with 98 labels for classification, our
Japanese-based language model achieves accuracy comparable to previous
research. This study extends the applicability of multimodal machine learning
frameworks to broader clinical studies and non-English languages.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:20:44 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Takahashi",
"Junichiro",
""
],
[
"Guan",
"JingChuan",
""
],
[
"Sato",
"Masataka",
""
],
[
"Baba",
"Kaito",
""
],
[
"Haruguchi",
"Kazuto",
""
],
[
"Nagashima",
"Daichi",
""
],
[
"Kodera",
"Satoshi",
""
],
[
"Takeda",
"Norihiko",
""
]
] |
2504.09303 | Martin Hardcastle | M. J. Hardcastle, J. C. S. Pierce, K. J. Duncan, G. G\"urkan, Y. Gong,
M. A. Horton, B. Mingo, H. J. A. R\"ottgering and D. J. B. Smith | Radio AGN selection in LoTSS DR2 | 23 pages, accepted for publication in MNRAS | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The wide-area component of the LOFAR Two-Metre Sky Survey (LoTSS) is
currently the largest radio survey ever carried out, and a large fraction of
the 4.5 million radio sources it contains have been optically identified with
galaxies or quasars with spectroscopic or photometric redshifts. Identification
of radio-luminous AGN from this LoTSS source catalogue is not only important
from the point of view of understanding the accretion history of the universe,
but also enables a wide range of other science. However, at present the vast
majority of the optical identifications lack spectroscopic information or
well-sampled spectral energy distributions. We show that colour and absolute
magnitude information from the Wide-Field Infrared Survey Explorer (WISE)
allows for the robust and efficient selection of radio AGN candidates,
generating a radio AGN candidate sample of around 600,000 objects with flux
density $> 1.1$ mJy, spanning 144-MHz luminosities between $10^{21}$ and
$10^{29}$ W Hz$^{-1}$. We use the catalogue to constrain the total sky density
of radio-luminous AGN and the evolution of their luminosity function between
$z=0$ and $z\approx 1$, and show that the typical mass of their host galaxies,
around $10^{11} M_\odot$, is essentially independent of radio luminosity above
around $L_{144} \approx 10^{24}$ W Hz$^{-1}$. Combining with Very Large Array
Sky Survey (VLASS) data, we show that the core prominences, radio spectral
indices and variability of extended sources from the sample are qualitatively
consistent with the expectations from unified models. A catalogue of the radio
AGN candidates is released with this paper.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:24:09 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hardcastle",
"M. J.",
""
],
[
"Pierce",
"J. C. S.",
""
],
[
"Duncan",
"K. J.",
""
],
[
"Gürkan",
"G.",
""
],
[
"Gong",
"Y.",
""
],
[
"Horton",
"M. A.",
""
],
[
"Mingo",
"B.",
""
],
[
"Röttgering",
"H. J. A.",
""
],
[
"Smith",
"D. J. B.",
""
]
] |
2504.09304 | Mikhail Volkov | Romain Gervalle and Mikhail S. Volkov | Black holes with electroweak hair -- the detailed derivation | 99 pages, many figures, many appendices | null | null | null | hep-th astro-ph.HE gr-qc hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a very detailed derivation of solutions describing hairy black
holes within the gravity-coupled Weinberg-Salam theory, which were previously
reported in
\href{https://doi.org/10.1103/PhysRevLett.133.171402}{Phys.Rev.Lett. 133 (2024)
171402}. These black holes support a strong magnetic field that polarizes the
electroweak vacuum and creates a condensate of massive fields carrying
superconducting currents along the black hole horizon. The currents, in turn,
generate a ``corona'' of magnetic vortex segments attached to the horizon at
both ends. The condensate and corona together constitute the black hole hair.
The extremal solutions approach, in the far field, the magnetic
Reissner-Nordstr\"om configuration, with a total mass that is {\it lower} than
the total charge, $M<|Q|$, due to the negative Zeeman energy of the condensate.
This makes the removal of the hair energetically unfavorable. The maximally
hairy black holes exhibit masses comparable to terrestrial values, with
approximately 11\% of their total mass stored in the hair. Given that these
solutions arise within a well-tested theoretical framework, they are likely to
have physical relevance.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:32:16 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Gervalle",
"Romain",
""
],
[
"Volkov",
"Mikhail S.",
""
]
] |
2504.09305 | Chee Ng | Owen Patterson, Chee Ng | Enhancing Contrastive Demonstration Selection with Semantic Diversity
for Robust In-Context Machine Translation | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In-Context Learning (ICL) empowers large language models to perform tasks by
conditioning on a few input-output examples. However, the performance of ICL is
highly sensitive to the selection of these demonstrations. While existing
methods focus on similarity or contrastive selection, they often overlook the
importance of diversity among the chosen examples. In this paper, we propose
DiverseConE (Diversity-Enhanced Contrastive Example Selection), a novel
approach for demonstration selection in in-context learning for machine
translation. Our method builds upon contrastive selection by incorporating a
diversity enhancement step based on embedding space dissimilarity. We conduct
extensive experiments on the Llama2-7b model across four language pairs
(English-Chinese, Chinese-English, Russian-German, German-Russian) in 1-shot
and 3-shot settings, using COMET20 and COMET22 for evaluation. Our results
demonstrate that DiverseConE consistently outperforms strong baseline methods,
including random selection, BM25, TopK, and a state-of-the-art contrastive
selection method. Further analysis, including diversity metrics and human
evaluation, validates the effectiveness of our approach and highlights the
benefits of considering demonstration diversity for improved translation
quality.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:35:04 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Patterson",
"Owen",
""
],
[
"Ng",
"Chee",
""
]
] |
2504.09306 | Paolo Caldiroli | Elvise Berchio and Paolo Caldiroli | Weighted Hardy-Rellich inequalities via the Emden-Fowler transform | null | null | null | null | math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We exploit a technique based on the Emden-Fowler transform to prove optimal
Hardy-Rellich inequalities on cones, including the punctured space
$\mathb{R}^N\setminus\{0\}$ and the half space as particular cases. We find
optimal constants for classes of test functions vanishing on the boundary of
the cone and possibly orthogonal to prescribed eigenspaces of the Laplace
Beltrami operator restricted to the spherical projection of the cone.
Furthermore, we show that extremals do not exist in the natural function
spaces. Depending on the parameters, certain resonance phenomena can occur. For
proper cones, this is excluded when considering test functions with compact
support. Finally, for suitable subsets of the cones we provide improved
Hardy-Rellich inequalities, under different boundary conditions, with optimal
remainder terms.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:39:16 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Berchio",
"Elvise",
""
],
[
"Caldiroli",
"Paolo",
""
]
] |
2504.09307 | Mingyu Liang | Mingyu Liang, Hiwot Tadese Kassa, Wenyin Fu, Brian Coutinho, Louis
Feng, Christina Delimitrou | Lumos: Efficient Performance Modeling and Estimation for Large-scale LLM
Training | Accepted to MLSys 2025 | null | null | null | cs.DC cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Training LLMs in distributed environments presents significant challenges due
to the complexity of model execution, deployment systems, and the vast space of
configurable strategies. Although various optimization techniques exist,
achieving high efficiency in practice remains difficult. Accurate performance
models that effectively characterize and predict a model's behavior are
essential for guiding optimization efforts and system-level studies. We propose
Lumos, a trace-driven performance modeling and estimation toolkit for
large-scale LLM training, designed to accurately capture and predict the
execution behaviors of modern LLMs. We evaluate Lumos on a production ML
cluster with up to 512 NVIDIA H100 GPUs using various GPT-3 variants,
demonstrating that it can replay execution time with an average error of just
3.3%, along with other runtime details, across different models and
configurations. Additionally, we validate its ability to estimate performance
for new setups from existing traces, facilitating efficient exploration of
model and deployment configurations.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:43:24 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Liang",
"Mingyu",
""
],
[
"Kassa",
"Hiwot Tadese",
""
],
[
"Fu",
"Wenyin",
""
],
[
"Coutinho",
"Brian",
""
],
[
"Feng",
"Louis",
""
],
[
"Delimitrou",
"Christina",
""
]
] |
2504.09308 | Adnan Hajomer Dr. | Adnan A.E. Hajomer, Axl Bomhals, C\'Edric Bruynsteen, Aboobackkar
Sidhique, Ivan Derkach, Ulrik L. Andersen, Xin Yin, and Tobias Gehring | Chip-Based 16 GBaud Continuous-Variable Quantum Key Distribution | null | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Quantum key distribution (QKD) stands as the most successful application of
quantum information science, providing information-theoretic security for key
exchange. While it has evolved from proof-of-concept experiments to commercial
products, widespread adoption requires chip-based integration to reduce costs,
enable mass production, facilitate miniaturization, and enhance system
performance. Here, we demonstrate the first fully photonic-integrated
continuous-variable QKD (CVQKD) system operating at a classical telecom symbol
rate of 16 GBaud. Our system integrates a silicon photonic transmitter circuit
(excluding the laser source) and a 20 GHz photonic-electronic receiver, which
features a phase-diverse silicon photonic integrated circuit and
custom-designed GaAs pHEMT transimpedance amplifiers. Advanced digital signal
processing allows our system to achieve the highest reported secure key rate to
date, reaching 0.289 Gb/s and 0.246 Gb/s over a 20 km fiber link in the
asymptotic and finite-size regimes, respectively. These results establish a
record key rate and represent a critical step toward scalable, cost-effective,
and mass-deployable quantum-secure communication using photonic-integrated
CVQKD systems.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:55:52 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hajomer",
"Adnan A. E.",
""
],
[
"Bomhals",
"Axl",
""
],
[
"Bruynsteen",
"CÉdric",
""
],
[
"Sidhique",
"Aboobackkar",
""
],
[
"Derkach",
"Ivan",
""
],
[
"Andersen",
"Ulrik L.",
""
],
[
"Yin",
"Xin",
""
],
[
"Gehring",
"Tobias",
""
]
] |
2504.09309 | Emily Johnson | Emily Johnson, Xavier Holt, Noah Wilson | Improving the Accuracy and Efficiency of Legal Document Tagging with
Large Language Models and Instruction Prompts | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Legal multi-label classification is a critical task for organizing and
accessing the vast amount of legal documentation. Despite its importance, it
faces challenges such as the complexity of legal language, intricate label
dependencies, and significant label imbalance. In this paper, we propose
Legal-LLM, a novel approach that leverages the instruction-following
capabilities of Large Language Models (LLMs) through fine-tuning. We reframe
the multi-label classification task as a structured generation problem,
instructing the LLM to directly output the relevant legal categories for a
given document. We evaluate our method on two benchmark datasets, POSTURE50K
and EURLEX57K, using micro-F1 and macro-F1 scores. Our experimental results
demonstrate that Legal-LLM outperforms a range of strong baseline models,
including traditional methods and other Transformer-based approaches.
Furthermore, ablation studies and human evaluations validate the effectiveness
of our approach, particularly in handling label imbalance and generating
relevant and accurate legal labels.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:57:04 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Johnson",
"Emily",
""
],
[
"Holt",
"Xavier",
""
],
[
"Wilson",
"Noah",
""
]
] |
2504.09311 | Jiaxin Jiang | Jiaxin Jiang, Siyuan Yao, Yuchen Li, Qiange Wang, Bingsheng He, Min
Chen | Dupin: A Parallel Framework for Densest Subgraph Discovery in Fraud
Detection on Massive Graphs (Technical Report) | null | null | null | null | cs.DB | http://creativecommons.org/licenses/by/4.0/ | Detecting fraudulent activities in financial and e-commerce transaction
networks is crucial. One effective method for this is Densest Subgraph
Discovery (DSD). However, deploying DSD methods in production systems faces
substantial scalability challenges due to the predominantly sequential nature
of existing methods, which impedes their ability to handle large-scale
transaction networks and results in significant detection delays. To address
these challenges, we introduce Dupin, a novel parallel processing framework
designed for efficient DSD processing in billion-scale graphs. Dupin is powered
by a processing engine that exploits the unique properties of the peeling
process, with theoretical guarantees on detection quality and efficiency. Dupin
provides userfriendly APIs for flexible customization of DSD objectives and
ensures robust adaptability to diverse fraud detection scenarios. Empirical
evaluations demonstrate that Dupin consistently outperforms several existing
DSD methods, achieving performance improvements of up to 100 times compared to
traditional approaches. On billion-scale graphs, Dupin demonstrates the
potential to enhance the prevention of fraudulent transactions from 45% to
94.5% and reduces density error from 30% to below 5%, as supported by our
experimental results. These findings highlight the effectiveness of Dupin in
real-world applications, ensuring both speed and accuracy in fraud detection.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:14:54 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Jiang",
"Jiaxin",
""
],
[
"Yao",
"Siyuan",
""
],
[
"Li",
"Yuchen",
""
],
[
"Wang",
"Qiange",
""
],
[
"He",
"Bingsheng",
""
],
[
"Chen",
"Min",
""
]
] |
2504.09312 | William Pires | Xi Chen, William Pires, Toniann Pitassi, Rocco A. Servedio | Testing Juntas and Junta Subclasses with Relative Error | null | null | null | null | cs.CC cs.DM cs.DS | http://creativecommons.org/licenses/by/4.0/ | This papers considers the junta testing problem in a recently introduced
``relative error'' variant of the standard Boolean function property testing
model. In relative-error testing we measure the distance from $f$ to $g$, where
$f,g: \{0,1\}^n \to \{0,1\}$, by the ratio of $|f^{-1}(1) \triangle g^{-1}(1)|$
(the number of inputs on which $f$ and $g$ disagree) to $|f^{-1}(1)|$ (the
number of satisfying assignments of $f$), and we give the testing algorithm
both black-box access to $f$ and also access to independent uniform samples
from $f^{-1}(1)$.
Chen et al. (SODA 2025) observed that the class of $k$-juntas is
$\text{poly}(2^k,1/\epsilon)$-query testable in the relative-error model, and
asked whether $\text{poly}(k,1/\epsilon)$ queries is achievable. We answer this
question affirmatively by giving a $\tilde{O}(k/\epsilon)$-query algorithm,
matching the optimal complexity achieved in the less challenging standard
model. Moreover, as our main result, we show that any subclass of $k$-juntas
that is closed under permuting variables is relative-error testable with a
similar complexity. This gives highly efficient relative-error testing
algorithms for a number of well-studied function classes, including size-$k$
decision trees, size-$k$ branching programs, and size-$k$ Boolean formulas.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:17:41 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Chen",
"Xi",
""
],
[
"Pires",
"William",
""
],
[
"Pitassi",
"Toniann",
""
],
[
"Servedio",
"Rocco A.",
""
]
] |
2504.09313 | Subramaniyan Sabari | S. Sabari, R. Sasireka, R. Radha, A. Uthayakumar, and L. Tomio | Vortices in Tunable Dipolar Bose-Einstein condensates with Attractive
Interactions | 10 pages, 8 figures | null | null | null | cond-mat.quant-gas | http://creativecommons.org/licenses/by/4.0/ | We investigate the formation of vortices in quasi-two-dimensional dipolar
Bose-Einstein Condensates (BECs) through the interplay between two-body contact
and long-ranged dipole-dipole interactions (DDIs), as both interactions can be
tuned from repulsive to attractive. By solving the associated Gross-Pitaevskii
equation for a rotating system, our initial approach concentrates on
stabilizing a collapsing condensate with attractive s-wave two-body
interactions by employing sufficiently large repulsive DDIs. Subsequently, the
same procedure was applied after reversing the signs of both interactions to
evaluate the sensitivity of vortex formation to such an interchange of
interactions. As a reference to guide our investigation, valid for generic
dipolar atomic species, we have assumed a condensate with the strong dipolar
dysprosium isotope, 164Dy. The correlation of the results with other dipolar
BEC systems was exemplified by considering rotating BECs with two other
isotopes, namely 168Er and 52Cr. For a purely dipolar condensate (with zero
contact interactions) under fixed rotation, we demonstrate how the number of
visible vortices increases as the DDI becomes more repulsive, accomplished by
tuning the orientation of the dipoles through a characteristic angle parameter.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:23:19 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Sabari",
"S.",
""
],
[
"Sasireka",
"R.",
""
],
[
"Radha",
"R.",
""
],
[
"Uthayakumar",
"A.",
""
],
[
"Tomio",
"L.",
""
]
] |
2504.09314 | Sonja Klisch | Sonja Klisch | Leading soft theorems on plane wave backgrounds | 42+4 pages, 5 figures | null | null | null | hep-th gr-qc | http://creativecommons.org/licenses/by/4.0/ | The infrared singularities of scattering amplitudes have historically
contributed to much development in understanding fundamental structures in
physics. However, the fate of the leading soft singularities of amplitudes in
non-trivial background fields has remained largely unknown. In this paper, we
derive the leading soft theorems for photons, gluons and gravitons on generic
plane wave backgrounds in gauge theory and gravity. The results differ from the
flat space results through dependence on the initial conditions of the soft
mediator. We also consider the special case of self-dual plane wave
backgrounds, and match onto the flat space results when the background is
treated perturbatively.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:23:53 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Klisch",
"Sonja",
""
]
] |
2504.09315 | Faisal Haque Bappy | Tahrim Hossain, Faisal Haque Bappy, Tarannum Shaila Zaman, Raiful
Hasan, Tariqul Islam | SmartShift: A Secure and Efficient Approach to Smart Contract Migration | null | null | null | null | cs.CR cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Blockchain and smart contracts have emerged as revolutionary technologies
transforming distributed computing. While platform evolution and smart
contracts' inherent immutability necessitate migrations both across and within
chains, migrating the vast amounts of critical data in these contracts while
maintaining data integrity and minimizing operational disruption presents a
significant challenge. To address these challenges, we present SmartShift, a
framework that enables secure and efficient smart contract migrations through
intelligent state partitioning and progressive function activation, preserving
operational continuity during transitions. Our comprehensive evaluation
demonstrates that SmartShift significantly reduces migration downtime while
ensuring robust security, establishing a foundation for efficient and secure
smart contract migration systems.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:28:23 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hossain",
"Tahrim",
""
],
[
"Bappy",
"Faisal Haque",
""
],
[
"Zaman",
"Tarannum Shaila",
""
],
[
"Hasan",
"Raiful",
""
],
[
"Islam",
"Tariqul",
""
]
] |
2504.09316 | Raj Kumar Mistri Dr. | Raj Kumar Mistri, Nitesh Prajapati | Direct and Inverse Problems for Restricted Signed Sumsets -- I | 35 pages | null | null | null | math.NT math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Let $A=\{a_{1},\ldots,a_{k}\}$ be a nonempty finite subset of an additive
abelian group $G$. For a positive integer $h$, the $h$-fold signed sumset of
$A$, denoted by $h_{\pm}A$, is defined as $$h_{\pm}A=\left\lbrace
\sum_{i=1}^{k} \lambda_{i} a_{i}: \lambda_{i} \in \{-h, \ldots, 0, \ldots, h\}
\ \text{for} \ i= 1, 2, \ldots, k \ \text{and} \ \sum_{i=1}^{k}
\left|\lambda_{i} \right| =h\right\rbrace,$$ and the restricted $h$-fold signed
sumset of $A$, denoted by $h^{\wedge}_{\pm}A$, is defined as
$$h^{\wedge}_{\pm}A=\left\lbrace \sum_{i=1}^{k} \lambda_{i} a_{i}: \lambda_{i}
\in \left\lbrace -1, 0, 1\right\rbrace \ \text{for} \ i= 1, 2, \ldots, k \
\text{and} \ \sum_{i=1}^{k} \left|\lambda_{i} \right| = h\right\rbrace. $$ A
direct problem for the sumset $h^{\wedge}_{\pm}A$ is to find the optimal size
of $h^{\wedge}_{\pm}A$ in terms of $h$ and $|A|$. An inverse problem for this
sumset is to determine the structure of the underlying set $A$ when the sumset
$h^{\wedge}_{\pm}A$ has optimal size. While some results are known for the
signed sumsets in finite abelian groups due to Bajnok and Matzke, not much is
known for the restricted $h$-fold signed sumset $h^{\wedge}_{\pm}A$ even in the
additive group of integers $\Bbb Z$. In case of $G = \Bbb Z$, Bhanja, Komatsu
and Pandey studied these problems for the sumset $h^{\wedge}_{\pm}A$ for $h=2,
3$, and $k$, and conjectured the direct and inverse results for $h \geq 4$. In
this paper, we prove these conjectures completely for the sets of positive
integers. In a subsequent paper, we prove these conjectures for the sets of
nonnegative integers.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:33:51 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Mistri",
"Raj Kumar",
""
],
[
"Prajapati",
"Nitesh",
""
]
] |
2504.09317 | Gui Zhou | Gui Zhou and Vasilis Papanikolaou and Zhiguo Ding and Robert Schober | Channel Estimation for mmWave Pinching-Antenna Systems | null | null | null | null | eess.SP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The full potential of pinching-antenna systems (PAS) can be unblocked if
pinching antennas can be accurately activated at positions tailored for the
serving users', which means that acquiring accurate channel state information
(CSI) at arbitrary positions along the waveguide is essential for the precise
placement of antennas. In this work, we propose an innovative channel
estimation scheme for millimeter-wave (mmWave) PAS. The proposed approach
requires activating only a small number of pinching antennas, thereby limiting
antenna switching and pilot overhead. Specifically, a base station (BS)
equipped with a waveguide selectively activates subarrays located near and far
from the feed point, each comprising a small number of pinching antennas. This
configuration effectively emulates a large-aperture array, enabling
high-accuracy estimation of multipath propagation parameters, including angles,
delays, and path gains. Simulation results demonstrate that the proposed method
achieves accurate CSI estimation and data rates while effectively reducing
hardware switching and pilot overhead.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:34:15 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhou",
"Gui",
""
],
[
"Papanikolaou",
"Vasilis",
""
],
[
"Ding",
"Zhiguo",
""
],
[
"Schober",
"Robert",
""
]
] |
2504.09318 | Waldemir Cambiucci | Waldemir Cambiucci, Regina Melo Silveira, Wilson Vicente Ruggiero | Hypergraphic representation for adaptive quantum circuits | Accepted in QCNC 2025 Nara, Japan. 7 pages, 7 figures, 1 table, 3
algorithms | null | null | null | quant-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Adaptive quantum circuits enhance flexibility and efficiency over traditional
static circuits by dynamically adjusting their structure and parameters in
real-time based on intermediate measurement outcomes. This paper introduces a
novel hypergraph representation for adaptive quantum circuits, where groups of
gates are considered as participants of hyperedges. By incorporating these gate
groups into hyperedges, we create an extended hypergraph that includes
constraints usable during the partitioning process. This approach guides the
partitioning to maintain groups of ports associated with classical operations,
ensuring that the resulting partitions prioritize qubits involved in the same
sections of classical operations inherent to the adaptive approach. We present
a new hypergraph partitioning algorithm based as an extension of
Fiduccia-Mattheyses heuristic, to support hypergraphs created from adaptive
quantum circuits. Comparative analysis between static and adaptive methods
demonstrates the effectiveness of the proposed hypergraph techniques for
adaptive circuits. Experimental results using benchmark quantum circuits
validate our theoretical insights, showing improvements in circuit
representation for partitioning heuristics. These findings highlight the
practical benefits of hypergraph representation in adaptive quantum computing.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:40:09 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Cambiucci",
"Waldemir",
""
],
[
"Silveira",
"Regina Melo",
""
],
[
"Ruggiero",
"Wilson Vicente",
""
]
] |
2504.09319 | Faisal Haque Bappy | Tahrim Hossain, Faisal Haque Bappy, Tarannum Shaila Zaman, and Tariqul
Islam | CrossLink: A Decentralized Framework for Secure Cross-Chain Smart
Contract Execution | null | null | null | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper introduces CrossLink, a decentralized framework for secure
cross-chain smart contract execution that effectively addresses the inherent
limitations of contemporary solutions, which primarily focus on asset transfers
and rely on potentially vulnerable centralized intermediaries. Recognizing the
escalating demand for seamless interoperability among decentralized
applications, CrossLink provides a trustless mechanism for smart contracts
across disparate blockchain networks to communicate and interact. At its core,
CrossLink utilizes a compact chain for selectively storing authorized contract
states and employs a secure inter-chain messaging mechanism to ensure atomic
execution and data consistency. By implementing a deposit/collateral fee system
and efficient state synchronization, CrossLink enhances security and mitigates
vulnerabilities, offering a novel approach to seamless, secure, and
decentralized cross-chain interoperability. A formal security analysis further
validates CrossLink's robustness against unauthorized modifications and
denial-of-service attacks.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:41:57 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hossain",
"Tahrim",
""
],
[
"Bappy",
"Faisal Haque",
""
],
[
"Zaman",
"Tarannum Shaila",
""
],
[
"Islam",
"Tariqul",
""
]
] |
2504.09320 | Mohammad Ivaki | Yingxiang Hu, Mohammad N. Ivaki, Julian Scheuer | Capillary Christoffel-Minkowski problem | null | null | null | null | math.DG math.AP math.MG | http://creativecommons.org/licenses/by/4.0/ | The result of Guan and Ma (Invent. Math. 151 (2003)) states that if
$\phi^{-1/k} : \mathbb{S}^n \to (0,\infty)$ is spherically convex, then $\phi$
arises as the $\sigma_k$ curvature (the $k$-th elementary symmetric function of
the principal radii of curvature) of a strictly convex hypersurface. In this
paper, we establish an analogous result in the capillary setting in the
half-space for $\theta\in(0,\pi/2)$: if $\phi^{-1/k} : \mathcal{C}_{\theta} \to
(0,\infty)$ is a capillary function and spherically convex, then $\phi$ is the
$\sigma_k$ curvature of a strictly convex capillary hypersurface.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:47:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hu",
"Yingxiang",
""
],
[
"Ivaki",
"Mohammad N.",
""
],
[
"Scheuer",
"Julian",
""
]
] |
2504.09321 | Alexei Koshelev E. | A. E. Koshelev | Directional driving of vortex lines with oscillating magnetic field | 9 pages, 7 figures, submitted to Phys. Rev. Applied | null | null | null | cond-mat.supr-con | http://creativecommons.org/licenses/by/4.0/ | The possibility of driving vortex lines with an oscillating magnetic field
could be useful in many applications. For example, it can be used for the
removal of undesired trapped flux from contactless elements of superconducting
devices. We investigate the dynamics of vortex lines in a superconducting film
with a ratchet thickness profile driven by an oscillating magnetic field
applied parallel to the film. We numerically simulate the dynamics of a single
flux line modeled as an elastic string with a variable length. We explore the
behavior for different frequencies and amplitudes of the oscillating magnetic
field and find several dynamic regimes. For moderate frequencies, the average
velocity is finite only within specific amplitude ranges. A notable feature is
the presence of extended velocity plateaus, which correspond to regimes when
the line moves by integer multiples of the spatial period $w$ during integer
multiples of the time period $T$. The transitions to these plateau states are
rather steep, especially at low frequencies. The plateau at velocity $w/T$
dominates at intermediate frequencies but vanishes at high frequencies. The
onset field amplitude of finite velocity nonmonotonically depends on the
frequency and passes through a minimum at a certain frequency value. At low
frequencies, the velocity exceeds $w/T$ and progressively increases with the
amplitude. These findings provide valuable insights into the dynamic behavior
of vortex lines driven by oscillating magnetic field in patterned
superconducting films, offering potential pathways for controlling the magnetic
flux in superconducting devices.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:50:50 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Koshelev",
"A. E.",
""
]
] |
2504.09322 | Tyler Spears | Tyler Spears, Shen Zhu, Yinzhu Jin, Aman Shrivastava, P. Thomas
Fletcher | MedIL: Implicit Latent Spaces for Generating Heterogeneous Medical
Images at Arbitrary Resolutions | null | null | null | null | cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | In this work, we introduce MedIL, a first-of-its-kind autoencoder built for
encoding medical images with heterogeneous sizes and resolutions for image
generation. Medical images are often large and heterogeneous, where fine
details are of vital clinical importance. Image properties change drastically
when considering acquisition equipment, patient demographics, and pathology,
making realistic medical image generation challenging. Recent work in latent
diffusion models (LDMs) has shown success in generating images resampled to a
fixed-size. However, this is a narrow subset of the resolutions native to image
acquisition, and resampling discards fine anatomical details. MedIL utilizes
implicit neural representations to treat images as continuous signals, where
encoding and decoding can be performed at arbitrary resolutions without prior
resampling. We quantitatively and qualitatively show how MedIL compresses and
preserves clinically-relevant features over large multi-site, multi-resolution
datasets of both T1w brain MRIs and lung CTs. We further demonstrate how MedIL
can influence the quality of images generated with a diffusion model, and
discuss how MedIL can enhance generative models to resemble raw clinical
acquisitions.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:52:56 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Spears",
"Tyler",
""
],
[
"Zhu",
"Shen",
""
],
[
"Jin",
"Yinzhu",
""
],
[
"Shrivastava",
"Aman",
""
],
[
"Fletcher",
"P. Thomas",
""
]
] |
2504.09324 | Daniil Lukin | Daniil M. Lukin, Bennet Windt, Miguel Bello, Dominic Catanzaro,
Melissa A. Guidry, Eran Lustig, Souvik Biswas, Giovanni Scuri, Trung Kien Le,
Joshua Yang, Arina A. Nikitina, Misagh Ghezellou, Hiroshi Abe, Takeshi
Ohshima, Jawad Ul-Hassan, and Jelena Vu\v{c}kovi\'c | Mesoscopic cavity quantum electrodynamics with phase-disordered emitters
in a Kerr nonlinear resonator | null | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | The field of cavity quantum electrodynamics (QED) has seen a recent
resurgence of interest in few- and many-body physics owing to the realization
that the breaking of symmetries and the presence of disorder can give rise to
entirely new phenomena. Here we demonstrate a few-emitter cavity QED system
capable of realizing new Hamiltonians in quantum optics based on breaking of
symmetries and the realization of an in situ Kerr nonlinearity. Our experiment
relies on a high-finesse silicon carbide whispering gallery mode resonator
hosting an ensemble of silicon vacancy color centers. The simultaneous presence
of spectral and spatial disorder of the mesoscopic atom system gives rise to
emergent chirality, and the optical nonlinearity of the silicon carbide host
crystal enables the observation of atom-photon correlations induced by a
four-photon nonlinear process. This work demonstrates the potential for solid
state defect systems to realize emerging proposals and to study fundamental
physics in quantum electrodynamics.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 19:55:06 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Lukin",
"Daniil M.",
""
],
[
"Windt",
"Bennet",
""
],
[
"Bello",
"Miguel",
""
],
[
"Catanzaro",
"Dominic",
""
],
[
"Guidry",
"Melissa A.",
""
],
[
"Lustig",
"Eran",
""
],
[
"Biswas",
"Souvik",
""
],
[
"Scuri",
"Giovanni",
""
],
[
"Le",
"Trung Kien",
""
],
[
"Yang",
"Joshua",
""
],
[
"Nikitina",
"Arina A.",
""
],
[
"Ghezellou",
"Misagh",
""
],
[
"Abe",
"Hiroshi",
""
],
[
"Ohshima",
"Takeshi",
""
],
[
"Ul-Hassan",
"Jawad",
""
],
[
"Vučković",
"Jelena",
""
]
] |
2504.09325 | Muneer Al-Zubi | Muneer M. Al-Zubi and Mohamed-Slim Alouini | Macroscale Molecular Communication in IoT-based Pipeline Inspection and
Monitoring Applications: Preliminary Experiment and Mathematical Model | null | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | Today, pipeline networks serve as critical infrastructure for transporting
materials such as water, gas, and oil. Modern technologies such as the Internet
of Things (IoT), sensor nodes, and inspection robots enable efficient pipeline
monitoring and inspection. They can help detect and monitor various conditions
and defects in pipelines such as cracks, corrosion, leakage, pressure, flow,
and temperature. Since most pipelines are buried underground, wireless
communication links suffer from significant attenuation and noise due to harsh
environmental conditions. In such systems, communication links are required
between the sensor nodes as well as between the external control/monitoring
unit or sensor node and the inspection robot inside the pipeline. In this
paper, we propose a macroscale molecular communication (MC) system in the
IoT-based pipeline inspection and monitoring networks to address this
challenge. We develop a mathematical model and implement a preliminary
experimental testbed to validate the system and demonstrate its feasibility by
transmitting and reconstructing binary sequences using volatile organic
compound (VOC) as an information signal. We examined the impact of various
system parameters including airflow carrier velocity, released VOC velocity,
emission duration, and bit duration. Results indicate that these parameters
significantly influence the received molecular signal, emphasizing the need for
optimal configuration. This work serves as a preliminary step for further
research on the application of MC in IoT-based pipeline inspection and
monitoring systems.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:00:12 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Al-Zubi",
"Muneer M.",
""
],
[
"Alouini",
"Mohamed-Slim",
""
]
] |
2504.09326 | Huai Qian Khor Mr. | Huai-Qian Khor, Yante Li, Xingxun Jiang, Guoying Zhao | Infused Suppression Of Magnification Artefacts For Micro-AU Detection | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Facial micro-expressions are spontaneous, brief and subtle facial motions
that unveil the underlying, suppressed emotions. Detecting Action Units (AUs)
in micro-expressions is crucial because it yields a finer representation of
facial motions than categorical emotions, effectively resolving the ambiguity
among different expressions. One of the difficulties in micro-expression
analysis is that facial motions are subtle and brief, thereby increasing the
difficulty in correlating facial motion features to AU occurrence. To bridge
the subtlety issue, flow-related features and motion magnification are a few
common approaches as they can yield descriptive motion changes and increased
motion amplitude respectively. While motion magnification can amplify the
motion changes, it also accounts for illumination changes and projection errors
during the amplification process, thereby creating motion artefacts that
confuse the model to learn inauthentic magnified motion features. The problem
is further aggravated in the context of a more complicated task where more AU
classes are analyzed in cross-database settings. To address this issue, we
propose InfuseNet, a layer-wise unitary feature infusion framework that
leverages motion context to constrain the Action Unit (AU) learning within an
informative facial movement region, thereby alleviating the influence of
magnification artefacts. On top of that, we propose leveraging magnified latent
features instead of reconstructing magnified samples to limit the distortion
and artefacts caused by the projection inaccuracy in the motion reconstruction
process. Via alleviating the magnification artefacts, InfuseNet has surpassed
the state-of-the-art results in the CD6ME protocol. Further quantitative
studies have also demonstrated the efficacy of motion artefacts alleviation.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:08:21 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Khor",
"Huai-Qian",
""
],
[
"Li",
"Yante",
""
],
[
"Jiang",
"Xingxun",
""
],
[
"Zhao",
"Guoying",
""
]
] |
2504.09327 | Andre LeClair | Andr\'e LeClair | Non-perturbative renormalization group for Higgs-like models in 4D | 23 pages, 13 figures | null | null | null | hep-th hep-ph | http://creativecommons.org/licenses/by/4.0/ | We recently defined a model of 2 coupled SU(2) Higgs doublets which revealed
an interesting structure of renormalization group flows to 1-loop when the
SU(2) is broken to U(1). In this article we compute the beta functions to 3
loops and show that the 1-loop structure of flows persists to higher orders.
For SU(2) broken to U(1), we conjecture a beta function to all orders. The
flows can be extended to large coupling using a strong-weak coupling duality $g
\to 1/g$. One finds a line of fixed points which are new conformal field
theories in 4 spacetime dimensions which are non-unitary due to negative norm
states but still have real eigenvalues. We also find massless flows between 2
non-trivial fixed points, and a regime with a cyclic RG flow. For the flows
between points on the critical line, we compute the anomalous dimensions of the
perturbations in the UV and IR, and identify some special points where
anomalous dimensions are rational numbers. The model is non-unitary since the
hamiltonian is pseudo-hermitian, $H^\dagger = {\cal K} H {\cal K}^\dagger$. The
unitary operator ${\cal K}$ satisfies ${\cal K}^2 =1$ and this allows a
projection onto positive norm states with a unitary time evolution with
positive probabilities.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:10:51 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"LeClair",
"André",
""
]
] |
2504.09328 | Sonia Laguna | Sonia Laguna, Alberto Garcia-Garcia, Marie-Julie Rakotosaona,
Stylianos Moschoglou, Leonhard Helminger, Sergio Orts-Escolano | Text To 3D Object Generation For Scalable Room Assembly | Published at the ICLR 2025 Workshop on Synthetic Data | null | null | null | cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Modern machine learning models for scene understanding, such as depth
estimation and object tracking, rely on large, high-quality datasets that mimic
real-world deployment scenarios. To address data scarcity, we propose an
end-to-end system for synthetic data generation for scalable, high-quality, and
customizable 3D indoor scenes. By integrating and adapting text-to-image and
multi-view diffusion models with Neural Radiance Field-based meshing, this
system generates highfidelity 3D object assets from text prompts and
incorporates them into pre-defined floor plans using a rendering tool. By
introducing novel loss functions and training strategies into existing methods,
the system supports on-demand scene generation, aiming to alleviate the
scarcity of current available data, generally manually crafted by artists. This
system advances the role of synthetic data in addressing machine learning
training limitations, enabling more robust and generalizable models for
real-world applications.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:13:07 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Laguna",
"Sonia",
""
],
[
"Garcia-Garcia",
"Alberto",
""
],
[
"Rakotosaona",
"Marie-Julie",
""
],
[
"Moschoglou",
"Stylianos",
""
],
[
"Helminger",
"Leonhard",
""
],
[
"Orts-Escolano",
"Sergio",
""
]
] |
2504.09329 | Giovane Galindo | Eder M. Correa, Giovane Galindo, Lino Grama | Chern-Ricci flow and t-Gauduchon Ricci-flat condition | 26 pages | null | null | null | math.DG math.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we study the $t$-Gauduchon Ricci-flat condition under the
Chern-Ricci flow. In this setting, we provide examples of Chern-Ricci flow on
compact non-K\"ahler Calabi-Yau manifolds which do not preserve the
$t$-Gauduchon Ricci-flat condition for $t<1$. The approach presented
generalizes some previous constructions on Hopf manifolds. Also, we provide
non-trivial new examples of balanced non-pluriclosed solution to the
pluriclosed flow on non-K\"ahler manifolds. Further, we describe the limiting
behavior, in the Gromov-Hausdorff sense, of geometric flows of Hermitian
metrics (including the Chern-Ricci flow and the pluriclosed flow) on certain
principal torus bundles over flag manifolds. In this last setting, we describe
explicitly the Gromov-Hausdorff limit of the pluriclosed flow on principal
$T^{2}$-bundles over the Fano threefold ${\mathbb{P}}(T_{{\mathbb{P}^{2}}})$.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:20:14 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Correa",
"Eder M.",
""
],
[
"Galindo",
"Giovane",
""
],
[
"Grama",
"Lino",
""
]
] |
2504.09330 | Sujay Nagaraj | Sujay Nagaraj, Yang Liu, Flavio P. Calmon, Berk Ustun | Regretful Decisions under Label Noise | The Thirteenth International Conference on Learning Representations
(ICLR 2025) | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Machine learning models are routinely used to support decisions that affect
individuals -- be it to screen a patient for a serious illness or to gauge
their response to treatment. In these tasks, we are limited to learning models
from datasets with noisy labels. In this paper, we study the instance-level
impact of learning under label noise. We introduce a notion of regret for this
regime which measures the number of unforeseen mistakes due to noisy labels. We
show that standard approaches to learning under label noise can return models
that perform well at a population level while subjecting individuals to a
lottery of mistakes. We present a versatile approach to estimate the likelihood
of mistakes at the individual level from a noisy dataset by training models
over plausible realizations of datasets without label noise. This is supported
by a comprehensive empirical study of label noise in clinical prediction tasks.
Our results reveal how failure to anticipate mistakes can compromise model
reliability and adoption, and demonstrate how we can address these challenges
by anticipating and avoiding regretful decisions.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:21:24 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Nagaraj",
"Sujay",
""
],
[
"Liu",
"Yang",
""
],
[
"Calmon",
"Flavio P.",
""
],
[
"Ustun",
"Berk",
""
]
] |
2504.09331 | Neekon Vafa | Andrej Bogdanov, Alon Rosen, Neekon Vafa, Vinod Vaikuntanathan | Adaptive Robustness of Hypergrid Johnson-Lindenstrauss | null | null | null | null | stat.CO cs.CC cs.DS | http://creativecommons.org/licenses/by/4.0/ | Johnson and Lindenstrauss (Contemporary Mathematics, 1984) showed that for $n
> m$, a scaled random projection $\mathbf{A}$ from $\mathbb{R}^n$ to
$\mathbb{R}^m$ is an approximate isometry on any set $S$ of size at most
exponential in $m$. If $S$ is larger, however, its points can contract
arbitrarily under $\mathbf{A}$. In particular, the hypergrid $([-B, B] \cap
\mathbb{Z})^n$ is expected to contain a point that is contracted by a factor of
$\kappa_{\mathsf{stat}} = \Theta(B)^{-1/\alpha}$, where $\alpha = m/n$.
We give evidence that finding such a point exhibits a
statistical-computational gap precisely up to $\kappa_{\mathsf{comp}} =
\widetilde{\Theta}(\sqrt{\alpha}/B)$. On the algorithmic side, we design an
online algorithm achieving $\kappa_{\mathsf{comp}}$, inspired by a discrepancy
minimization algorithm of Bansal and Spencer (Random Structures & Algorithms,
2020). On the hardness side, we show evidence via a multiple overlap gap
property (mOGP), which in particular captures online algorithms; and a
reduction-based lower bound, which shows hardness under standard worst-case
lattice assumptions.
As a cryptographic application, we show that the rounded
Johnson-Lindenstrauss embedding is a robust property-preserving hash function
(Boyle, Lavigne and Vaikuntanathan, TCC 2019) on the hypergrid for the
Euclidean metric in the computationally hard regime. Such hash functions
compress data while preserving $\ell_2$ distances between inputs up to some
distortion factor, with the guarantee that even knowing the hash function, no
computationally bounded adversary can find any pair of points that violates the
distortion bound.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:23:34 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Bogdanov",
"Andrej",
""
],
[
"Rosen",
"Alon",
""
],
[
"Vafa",
"Neekon",
""
],
[
"Vaikuntanathan",
"Vinod",
""
]
] |
2504.09332 | Cathy Mengying Fang | Cathy Mengying Fang, Wazeer Zulfikar, Yasith Samaradivakara, Suranga
Nanayakkara, Pattie Maes | The Goldilocks Time Window for Proactive Interventions in Wearable AI
Systems | null | null | null | null | cs.HC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | As AI systems become increasingly integrated into our daily lives and into
wearable form factors, there's a fundamental tension between their potential to
proactively assist us and the risk of creating intrusive, dependency-forming
experiences. This work proposes the concept of a Goldilocks Time Window -- a
contextually adaptive time window for proactive AI systems to deliver effective
interventions. We discuss the critical factors that determine the time window,
and the need of a framework for designing and evaluating proactive AI systems
that can navigate this tension successfully.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:30:03 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Fang",
"Cathy Mengying",
""
],
[
"Zulfikar",
"Wazeer",
""
],
[
"Samaradivakara",
"Yasith",
""
],
[
"Nanayakkara",
"Suranga",
""
],
[
"Maes",
"Pattie",
""
]
] |
2504.09333 | Robert Schippa | Sebastian Herr, Robert Schippa, Nikolay Tzvetkov | Global results for weakly dispersive KP-II equations on the cylinder | 63 pages | null | null | null | math.AP | http://creativecommons.org/licenses/by/4.0/ | We consider the dispersion-generalized KP-II equation on a partially periodic
domain in the weakly dispersive regime. We use Fourier decoupling techniques to
derive essentially sharp Strichartz estimates. With these at hand, we show
global well-posedness of the quasilinear Cauchy problem in $L^2(\mathbb{R}
\times \mathbb{T})$. Finally, we prove a long-time decay property of solutions
with small mass by using the Kato smoothing effect in the fractional case.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:30:24 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Herr",
"Sebastian",
""
],
[
"Schippa",
"Robert",
""
],
[
"Tzvetkov",
"Nikolay",
""
]
] |
2504.09334 | Waldemir Cambiucci | Waldemir Cambiucci, Regina Melo Silveira, Wilson Vicente Ruggiero | Spatial and temporal circuit cutting with hypergraphic partitioning | accepted in QuantumTech Summit-2025, International Experts Summit on
Quantum Technologies, Singapore 25 pages, 15 figures, 4 tables, 2 algorithms | null | null | null | quant-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Quantum computing promises to revolutionize problem-solving through quantum
mechanics, but current NISQ devices face limitations in qubit count and error
rates, hindering the execution of large-scale quantum circuits. To address
these challenges and improve scalability, two main circuit cutting strategies
have emerged: the gate-cut approach, which distributes circuit segments across
multiple QPUs (spatial), and the qubit wire cut approach, which divides
circuits for sequential execution (temporal). This paper presents a
hypergraph-based circuit cutting methodology suitable for both spatial and
temporal scenarios. By modeling quantum circuits as high-level hypergraphs, we
apply partitioning heuristics such as Stoer-Wagner, Fiduccia-Mattheyses, and
Kernighan-Lin to optimize the partitioning process. Our approach aims to reduce
communication overhead in spatial cuts and minimize qubit initialization costs
in temporal ones. To assess effectiveness, we propose a new evaluation metric
called the coupling ratio, which quantifies the trade-offs between
communication and initialization. Comparative analyses show that hypergraph
partitioning improves the efficiency of distributed quantum architectures.
Notably, the Fiduccia-Mattheyses heuristic offers superior performance and
adaptability for real-time circuit cutting on multi-QPU systems. Overall, this
work positions hypergraph partitioning as a foundational technique for scalable
quantum computing in distributed environments.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:31:07 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Cambiucci",
"Waldemir",
""
],
[
"Silveira",
"Regina Melo",
""
],
[
"Ruggiero",
"Wilson Vicente",
""
]
] |
2504.09335 | Jihoon Suh | Jihoon Suh, Takashi Tanaka | Efficient Implementation of Reinforcement Learning over Homomorphic
Encryption | 6 pages, 3 figures | Journal of The Society of Instrument and Control Engineers, vol.
64, no. 4, pp. 223-229, 2025 | 10.11499/sicejl.64.223 | null | cs.LG cs.CR cs.SY eess.SY | http://creativecommons.org/licenses/by/4.0/ | We investigate encrypted control policy synthesis over the cloud. While
encrypted control implementations have been studied previously, we focus on the
less explored paradigm of privacy-preserving control synthesis, which can
involve heavier computations ideal for cloud outsourcing. We classify control
policy synthesis into model-based, simulator-driven, and data-driven approaches
and examine their implementation over fully homomorphic encryption (FHE) for
privacy enhancements. A key challenge arises from comparison operations (min or
max) in standard reinforcement learning algorithms, which are difficult to
execute over encrypted data. This observation motivates our focus on
Relative-Entropy-regularized reinforcement learning (RL) problems, which
simplifies encrypted evaluation of synthesis algorithms due to their
comparison-free structures. We demonstrate how linearly solvable value
iteration, path integral control, and Z-learning can be readily implemented
over FHE. We conduct a case study of our approach through numerical simulations
of encrypted Z-learning in a grid world environment using the CKKS encryption
scheme, showing convergence with acceptable approximation error. Our work
suggests the potential for secure and efficient cloud-based reinforcement
learning.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:34:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Suh",
"Jihoon",
""
],
[
"Tanaka",
"Takashi",
""
]
] |
2504.09336 | Simon-Christian Klein | Simon-Christian Klein | Essentially Non-oscillatory Spectral Volume Methods | null | null | null | null | math.NA cs.NA math.AP | http://creativecommons.org/licenses/by/4.0/ | A new Essentially Non-oscillatory (ENO) recovery algorithm is developed and
tested in a Finite Volume method. The construction is hinged on a reformulation
of the reconstruction as the solution to a variational problem. The sign
property of the classical ENO algorithm is expressed as restrictions on the
admissible set of solutions to this variational problem. In conjunction with an
educated guessing algorithm for possible locations of discontinuities an ENO
reconstruction algorithm without divided differences or smoothness indicators
is constructed. No tunable parameters exist apart from the desired order and
stencil width. The desired order is in principle arbitrary, but growing
stencils are needed. While classical ENO methods consider all connected
stencils that surround a cell under consideration the proposed recovery method
uses a fixed stencil, simplifying efficient high order implementations.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:36:54 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Klein",
"Simon-Christian",
""
]
] |
2504.09337 | Mark Strikman | Leonid Frankfurt and Mark Strikman | Discovery of High-momentum Nucleon Correlations in Nuclei -- an Early
History | Submitted to Acta Physica Polonica B special volume dedicated to
Dmitry Diakonov, Victor Petrov and Maxim Polyakov | null | null | null | hep-ph nucl-th | http://creativecommons.org/publicdomain/zero/1.0/ | We summarize the reasons for using the light-cone mechanics of nuclei for
description of high-energy nuclear processes and describe the steps, which have
led to the discovery of high-momentum correlations in nuclei.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:39:30 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Frankfurt",
"Leonid",
""
],
[
"Strikman",
"Mark",
""
]
] |
2504.09338 | Marius J\"urgensen | Marius J\"urgensen, Jacob Steiner, Gil Refael, and Mikael C. Rechtsman | Multi-band fractional Thouless pumps | 6 pages | null | null | null | cond-mat.str-el cond-mat.other | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Quantization of particle transport lies at the heart of topological physics.
In Thouless pumps - dimensionally reduced versions of the integer quantum Hall
effect - quantization is dictated by the integer winding of single-band Wannier
states. Here, we show that repulsive interactions can drive a transition from
an integer- to a fractional-quantized Thouless pump (at fixed integer filling)
by stabilizing a crystal of multi-band Wannier states, each with fractional
winding. We numerically illustrate the concept in few-particle systems, and
show that a dynamical Hartree-Fock ansatz can quantitatively reproduce the
pumping phase diagram.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:42:56 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Jürgensen",
"Marius",
""
],
[
"Steiner",
"Jacob",
""
],
[
"Refael",
"Gil",
""
],
[
"Rechtsman",
"Mikael C.",
""
]
] |
2504.09339 | Sharan Sahu | Sharan Sahu | Towards Optimal Differentially Private Regret Bounds in Linear MDPs | 28 pages, 2 figures | null | null | null | cs.LG cs.DS stat.ML | http://creativecommons.org/licenses/by/4.0/ | We study regret minimization under privacy constraints in episodic
inhomogeneous linear Markov Decision Processes (MDPs), motivated by the growing
use of reinforcement learning (RL) in personalized decision-making systems that
rely on sensitive user data. In this setting, both transition probabilities and
reward functions are assumed to be linear in a feature mapping $\phi(s, a)$,
and we aim to ensure privacy through joint differential privacy (JDP), a
relaxation of differential privacy suited to online learning. Prior work has
established suboptimal regret bounds by privatizing the LSVI-UCB algorithm,
which achieves $\widetilde{O}(\sqrt{d^3 H^4 K})$ regret in the non-private
setting. Building on recent advances that improve this to minimax optimal
regret $\widetilde{O}(HD\sqrt{K})$ via LSVI-UCB++ with Bernstein-style bonuses,
we design a new differentially private algorithm by privatizing LSVI-UCB++ and
adapting techniques for variance-aware analysis from offline RL. Our algorithm
achieves a regret bound of $\widetilde{O}(d \sqrt{H^3 K} + H^{4.5} d^{7/6}
K^{1/2} / \epsilon)$, improving over previous private methods. Empirical
results show that our algorithm retains near-optimal utility compared to
non-private baselines, indicating that privacy can be achieved with minimal
performance degradation in this setting.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:51:51 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Sahu",
"Sharan",
""
]
] |
2504.09340 | Joel Oskarsson | Simon Adamov, Joel Oskarsson, Leif Denby, Tomas Landelius, Kasper
Hintz, Simon Christiansen, Irene Schicker, Carlos Osuna, Fredrik Lindsten,
Oliver Fuhrer, Sebastian Schemm | Building Machine Learning Limited Area Models: Kilometer-Scale Weather
Forecasting in Realistic Settings | 72 pages, 50 figures. Code is available at
https://github.com/joeloskarsson/neural-lam-dev/releases/tag/building-ml-lams | null | null | null | physics.ao-ph cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Machine learning is revolutionizing global weather forecasting, with models
that efficiently produce highly accurate forecasts. Apart from global
forecasting there is also a large value in high-resolution regional weather
forecasts, focusing on accurate simulations of the atmosphere for a limited
area. Initial attempts have been made to use machine learning for such limited
area scenarios, but these experiments do not consider realistic forecasting
settings and do not investigate the many design choices involved. We present a
framework for building kilometer-scale machine learning limited area models
with boundary conditions imposed through a flexible boundary forcing method.
This enables boundary conditions defined either from reanalysis or operational
forecast data. Our approach employs specialized graph constructions with
rectangular and triangular meshes, along with multi-step rollout training
strategies to improve temporal consistency. We perform systematic evaluation of
different design choices, including the boundary width, graph construction and
boundary forcing integration. Models are evaluated across both a Danish and a
Swiss domain, two regions that exhibit different orographical characteristics.
Verification is performed against both gridded analysis data and in-situ
observations, including a case study for the storm Ciara in February 2020. Both
models achieve skillful predictions across a wide range of variables, with our
Swiss model outperforming the numerical weather prediction baseline for key
surface variables. With their substantially lower computational cost, our
findings demonstrate great potential for machine learning limited area models
in the future of regional weather forecasting.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 20:58:02 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Adamov",
"Simon",
""
],
[
"Oskarsson",
"Joel",
""
],
[
"Denby",
"Leif",
""
],
[
"Landelius",
"Tomas",
""
],
[
"Hintz",
"Kasper",
""
],
[
"Christiansen",
"Simon",
""
],
[
"Schicker",
"Irene",
""
],
[
"Osuna",
"Carlos",
""
],
[
"Lindsten",
"Fredrik",
""
],
[
"Fuhrer",
"Oliver",
""
],
[
"Schemm",
"Sebastian",
""
]
] |
2504.09341 | Rafid Mahmood | Hsuan Wei Liao, Christopher Klugmann, Daniel Kondermann, Rafid Mahmood | Minority Reports: Balancing Cost and Quality in Ground Truth Data
Annotation | 39 pages | null | null | null | cs.LG cs.CV cs.HC | http://creativecommons.org/licenses/by/4.0/ | High-quality data annotation is an essential but laborious and costly aspect
of developing machine learning-based software. We explore the inherent tradeoff
between annotation accuracy and cost by detecting and removing minority reports
-- instances where annotators provide incorrect responses -- that indicate
unnecessary redundancy in task assignments. We propose an approach to prune
potentially redundant annotation task assignments before they are executed by
estimating the likelihood of an annotator disagreeing with the majority vote
for a given task. Our approach is informed by an empirical analysis over
computer vision datasets annotated by a professional data annotation platform,
which reveals that the likelihood of a minority report event is dependent
primarily on image ambiguity, worker variability, and worker fatigue.
Simulations over these datasets show that we can reduce the number of
annotations required by over 60% with a small compromise in label quality,
saving approximately 6.6 days-equivalent of labor. Our approach provides
annotation service platforms with a method to balance cost and dataset quality.
Machine learning practitioners can tailor annotation accuracy levels according
to specific application needs, thereby optimizing budget allocation while
maintaining the data quality necessary for critical settings like autonomous
driving technology.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 21:04:56 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Liao",
"Hsuan Wei",
""
],
[
"Klugmann",
"Christopher",
""
],
[
"Kondermann",
"Daniel",
""
],
[
"Mahmood",
"Rafid",
""
]
] |
2504.09342 | Ali Rasteh | Ali Rasteh, Sundeep Rangan | Computationally Efficient Signal Detection with Unknown Bandwidths | Submitted to the IEEE Open Journal of the Communications Society | null | null | null | eess.SP cs.SY eess.SY | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Signal detection in environments with unknown signal bandwidth and time
intervals is a basic problem in adversarial and spectrum-sharing scenarios.
This paper addresses the problem of detecting signals occupying unknown degrees
of freedom from non-coherent power measurements where the signal is constrained
to an interval in one dimension or hypercube in multiple dimensions. A
Generalized Likelihood Ratio Test (GLRT) is derived, resulting in a
straightforward metric involving normalized average signal energy on each
candidate signal set. We present bounds on false alarm and missed detection
probabilities, demonstrating their dependence on signal-to-noise ratios (SNR)
and signal set sizes. To overcome the inherent computational complexity of
exhaustive searches, we propose a computationally efficient binary search
method, reducing the complexity from O(N2) to O(N) for one-dimensional cases.
Simulations indicate that the method maintains performance near exhaustive
searches and achieves asymptotic consistency, with interval-of-overlap
converging to one under constant SNR as measurement size increases. The
simulation studies also demonstrate superior performance and reduced complexity
compared to contemporary neural network-based approaches, specifically
outperforming custom-trained U-Net models in spectrum detection tasks.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 21:06:56 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Rasteh",
"Ali",
""
],
[
"Rangan",
"Sundeep",
""
]
] |
2504.09343 | Yiran Du | Yiran Du | Confirmation Bias in Generative AI Chatbots: Mechanisms, Risks,
Mitigation Strategies, and Future Research Directions | null | null | null | null | cs.HC cs.AI | http://creativecommons.org/licenses/by/4.0/ | This article explores the phenomenon of confirmation bias in generative AI
chatbots, a relatively underexamined aspect of AI-human interaction. Drawing on
cognitive psychology and computational linguistics, it examines how
confirmation bias, commonly understood as the tendency to seek information that
aligns with existing beliefs, can be replicated and amplified by the design and
functioning of large language models. The article analyzes the mechanisms by
which confirmation bias may manifest in chatbot interactions, assesses the
ethical and practical risks associated with such bias, and proposes a range of
mitigation strategies. These include technical interventions, interface
redesign, and policy measures aimed at promoting balanced AI-generated
discourse. The article concludes by outlining future research directions,
emphasizing the need for interdisciplinary collaboration and empirical
evaluation to better understand and address confirmation bias in generative AI
systems.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 21:08:36 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Du",
"Yiran",
""
]
] |
2504.09344 | Tai An | Weiqiang Huang, Juecen Zhan, Yumeng Sun, Xu Han, Tai An, Nan Jiang | Context-Aware Adaptive Sampling for Intelligent Data Acquisition Systems
Using DQN | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multi-sensor systems are widely used in the Internet of Things, environmental
monitoring, and intelligent manufacturing. However, traditional fixed-frequency
sampling strategies often lead to severe data redundancy, high energy
consumption, and limited adaptability, failing to meet the dynamic sensing
needs of complex environments. To address these issues, this paper proposes a
DQN-based multi-sensor adaptive sampling optimization method. By leveraging a
reinforcement learning framework to learn the optimal sampling strategy, the
method balances data quality, energy consumption, and redundancy. We first
model the multi-sensor sampling task as a Markov Decision Process (MDP), then
employ a Deep Q-Network to optimize the sampling policy. Experiments on the
Intel Lab Data dataset confirm that, compared with fixed-frequency sampling,
threshold-triggered sampling, and other reinforcement learning approaches, DQN
significantly improves data quality while lowering average energy consumption
and redundancy rates. Moreover, in heterogeneous multi-sensor environments,
DQN-based adaptive sampling shows enhanced robustness, maintaining superior
data collection performance even in the presence of interference factors. These
findings demonstrate that DQN-based adaptive sampling can enhance overall data
acquisition efficiency in multi-sensor systems, providing a new solution for
efficient and intelligent sensing.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 21:11:56 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Huang",
"Weiqiang",
""
],
[
"Zhan",
"Juecen",
""
],
[
"Sun",
"Yumeng",
""
],
[
"Han",
"Xu",
""
],
[
"An",
"Tai",
""
],
[
"Jiang",
"Nan",
""
]
] |
2504.09345 | Yichao Yuan | Yichao Yuan, Lin Ma, Nishil Talati | MoE-Lens: Towards the Hardware Limit of High-Throughput MoE LLM Serving
Under Resource Constraints | null | null | null | null | cs.DC cs.AI | http://creativecommons.org/licenses/by/4.0/ | Mixture of Experts (MoE) LLMs, characterized by their sparse activation
patterns, offer a promising approach to scaling language models while avoiding
proportionally increasing the inference cost. However, their large parameter
sizes present deployment challenges in resource-constrained environments with
limited GPU memory capacity, as GPU memory is often insufficient to accommodate
the full set of model weights. Consequently, typical deployments rely on
CPU-GPU hybrid execution: the GPU handles compute-intensive GEMM operations,
while the CPU processes the relatively lightweight attention mechanism. This
setup introduces a key challenge: how to effectively optimize resource
utilization across CPU and GPU? Prior work has designed system optimizations
based on performance models with limited scope. Specifically, such models do
not capture the complex interactions between hardware properties and system
execution mechanisms. Therefore, previous approaches neither identify nor
achieve the hardware limit.
This paper presents MoE-Lens, a high-throughput MoE LLM inference system
designed through holistic performance modeling for resource-constrained
environments. Our performance model thoroughly analyzes various fundamental
system components, including CPU memory capacity, GPU compute power, and
workload characteristics, to understand the theoretical performance upper bound
of MoE inference. Furthermore, it captures the system execution mechanisms to
identify the key hardware bottlenecks and accurately predict the achievable
throughput. Informed by our performance model, MoE-Lens introduces an inference
system approaching hardware limits. Evaluated on diverse MoE models and
datasets, MoE-Lens outperforms the state-of-the-art solution by 4.6x on average
(up to 25.5x), with our theoretical model predicting performance with an
average 94% accuracy.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 21:26:56 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Yuan",
"Yichao",
""
],
[
"Ma",
"Lin",
""
],
[
"Talati",
"Nishil",
""
]
] |
2504.09346 | Shira Michel | Shira Michel, Sufi Kaur, Sarah Elizabeth Gillespie, Jeffrey Gleason,
Christo Wilson, Avijit Ghosh | "It's not a representation of me": Examining Accent Bias and Digital
Exclusion in Synthetic AI Voice Services | This paper has been accepted to FAccT 2025 | null | null | null | cs.HC cs.AI cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advances in artificial intelligence (AI) speech generation and voice
cloning technologies have produced naturalistic speech and accurate voice
replication, yet their influence on sociotechnical systems across diverse
accents and linguistic traits is not fully understood. This study evaluates two
synthetic AI voice services (Speechify and ElevenLabs) through a mixed methods
approach using surveys and interviews to assess technical performance and
uncover how users' lived experiences influence their perceptions of accent
variations in these speech technologies. Our findings reveal technical
performance disparities across five regional, English-language accents and
demonstrate how current speech generation technologies may inadvertently
reinforce linguistic privilege and accent-based discrimination, potentially
creating new forms of digital exclusion. Overall, our study highlights the need
for inclusive design and regulation by providing actionable insights for
developers, policymakers, and organizations to ensure equitable and socially
responsible AI speech technologies.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 21:31:22 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Michel",
"Shira",
""
],
[
"Kaur",
"Sufi",
""
],
[
"Gillespie",
"Sarah Elizabeth",
""
],
[
"Gleason",
"Jeffrey",
""
],
[
"Wilson",
"Christo",
""
],
[
"Ghosh",
"Avijit",
""
]
] |
2504.09349 | Yefeng Fan | Yefeng Fan and Simon White | Neural Posterior Estimation on Exponential Random Graph Models:
Evaluating Bias and Implementation Challenges | null | null | null | null | stat.ME stat.ML | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Exponential random graph models (ERGMs) are flexible probabilistic frameworks
to model statistical networks through a variety of network summary statistics.
Conventional Bayesian estimation for ERGMs involves iteratively exchanging with
an auxiliary variable due to the intractability of ERGMs, however, this
approach lacks scalability to large-scale implementations. Neural posterior
estimation (NPE) is a recent advancement in simulation-based inference, using a
neural network based density estimator to infer the posterior for models with
doubly intractable likelihoods for which simulations can be generated. While
NPE has been successfully adopted in various fields such as cosmology, little
research has investigated its use for ERGMs. Performing NPE on ERGM not only
provides a differing angle of resolving estimation for the intractable ERGM
likelihoods but also allows more efficient and scalable inference using the
amortisation properties of NPE, and therefore, we investigate how NPE can be
effectively implemented in ERGMs.
In this study, we present the first systematic implementation of NPE for
ERGMs, rigorously evaluating potential biases, interpreting the biases
magnitudes, and comparing NPE fittings against conventional Bayesian ERGM
fittings. More importantly, our work highlights ERGM-specific areas that may
impose particular challenges for the adoption of NPE.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 21:37:11 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Fan",
"Yefeng",
""
],
[
"White",
"Simon",
""
]
] |
2504.09350 | Mark Van Den Bosch | Mark van den Bosch and Hermen Jan Hupkes | Local Phase Tracking and Metastability of Planar Waves in Stochastic
Reaction-Diffusion Systems | 79 pages, 3 figures | null | null | null | math.AP math.DS math.PR | http://creativecommons.org/licenses/by/4.0/ | Planar travelling waves on $\mathbb R^d,$ with $ d\geq 2,$ are shown to
persist in systems of reaction-diffusion equations with multiplicative noise on
significantly long timescales with high probability, provided that the wave is
orbitally stable in dimension one ($d=1$). While a global phase tracking
mechanism is required to determine the location of the stochastically perturbed
wave in one dimension, or on a cylindrical domain, we show that the travelling
wave on the full unbounded space can be controlled by keeping track of local
deviations only. In particular, the energy infinitesimally added to or
withdrawn from the system by noise dissipates almost fully into the transverse
direction, leaving behind small localised phase shifts. The noise process
considered is white in time and coloured in space, possibly weighted, and
either translation invariant or trace class.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 21:44:34 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Bosch",
"Mark van den",
""
],
[
"Hupkes",
"Hermen Jan",
""
]
] |
2504.09351 | Arturo Mart\'inez-Celis | Arturo Mart\'inez-Celis, Adam Morawski | A small Radon-Nikod\'ym compact space from a parametrized diamond | 11 pages | null | null | null | math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A compact space $K$ is Radon-Nikod\'{y}m if there is a lower semi-continuous
metric fragmenting $K$. In this note, we show that, under $\diamondsuit
(\mathrm{non}{\mathcal{M}})$, there is a Radon-Nikod\'{y}m compact space of
weight $\aleph_1$ with a continuous image that is not Radon-Nikod\'{y}m, which
partially answers a question posed in arXiv:1112.4152 [math.FA].
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 21:49:52 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Martínez-Celis",
"Arturo",
""
],
[
"Morawski",
"Adam",
""
]
] |
2504.09352 | Iason Chaimalas | Iason Chaimalas, Arnas Vy\v{s}niauskas, Gabriel Brostow | Explorer: Robust Collection of Interactable GUI Elements | 19 pages, 17 figures | null | null | null | cs.HC cs.AI cs.CV | http://creativecommons.org/licenses/by/4.0/ | Automation of existing Graphical User Interfaces (GUIs) is important but hard
to achieve. Upstream of making the GUI user-accessible or somehow scriptable,
even the data-collection to understand the original interface poses significant
challenges. For example, large quantities of general UI data seem helpful for
training general machine learning (ML) models, but accessibility for each
person can hinge on the ML's precision on a specific app. We therefore take the
perspective that a given user needs confidence, that the relevant UI elements
are being detected correctly throughout one app or digital environment. We
mostly assume that the target application is known in advance, so that data
collection and ML-training can be personalized for the test-time target domain.
The proposed Explorer system focuses on detecting on-screen buttons and
text-entry fields, i.e. interactables, where the training process has access to
a live version of the application. The live application can run on almost any
popular platform except iOS phones, and the collection is especially
streamlined for Android phones or for desktop Chrome browsers. Explorer also
enables the recording of interactive user sessions, and subsequent mapping of
how such sessions overlap and sometimes loop back to similar states. We show
how having such a map enables a kind of path planning through the GUI, letting
a user issue audio commands to get to their destination. Critically, we are
releasing our code for Explorer openly at https://github.com/varnelis/Explorer.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 22:02:29 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Chaimalas",
"Iason",
""
],
[
"Vyšniauskas",
"Arnas",
""
],
[
"Brostow",
"Gabriel",
""
]
] |
2504.09353 | Mandeep Rathee | Mandeep Rathee, Venktesh V, Sean MacAvaney, Avishek Anand | Breaking the Lens of the Telescope: Online Relevance Estimation over
Large Retrieval Sets | 11 pages,5 figures, 4 tables | null | null | null | cs.IR | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Advanced relevance models, such as those that use large language models
(LLMs), provide highly accurate relevance estimations. However, their
computational costs make them infeasible for processing large document corpora.
To address this, retrieval systems often employ a telescoping approach, where
computationally efficient but less precise lexical and semantic retrievers
filter potential candidates for further ranking. However, this approach heavily
depends on the quality of early-stage retrieval, which can potentially exclude
relevant documents early in the process. In this work, we propose a novel
paradigm for re-ranking called online relevance estimation that continuously
updates relevance estimates for a query throughout the ranking process. Instead
of re-ranking a fixed set of top-k documents in a single step, online relevance
estimation iteratively re-scores smaller subsets of the most promising
documents while adjusting relevance scores for the remaining pool based on the
estimations from the final model using an online bandit-based algorithm. This
dynamic process mitigates the recall limitations of telescoping systems by
re-prioritizing documents initially deemed less relevant by earlier stages --
including those completely excluded by earlier-stage retrievers. We validate
our approach on TREC benchmarks under two scenarios: hybrid retrieval and
adaptive retrieval. Experimental results demonstrate that our method is
sample-efficient and significantly improves recall, highlighting the
effectiveness of our online relevance estimation framework for modern search
systems.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 22:05:50 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Rathee",
"Mandeep",
""
],
[
"V",
"Venktesh",
""
],
[
"MacAvaney",
"Sean",
""
],
[
"Anand",
"Avishek",
""
]
] |
2504.09354 | Duy-Cat Can | Duy-Cat Can, Quang-Huy Tang, Huong Ha, Binh T. Nguyen and Oliver Y.
Ch\'en | REMEMBER: Retrieval-based Explainable Multimodal Evidence-guided
Modeling for Brain Evaluation and Reasoning in Zero- and Few-shot
Neurodegenerative Diagnosis | null | null | null | null | cs.CV cs.AI cs.CL cs.LG q-bio.QM | http://creativecommons.org/licenses/by/4.0/ | Timely and accurate diagnosis of neurodegenerative disorders, such as
Alzheimer's disease, is central to disease management. Existing deep learning
models require large-scale annotated datasets and often function as "black
boxes". Additionally, datasets in clinical practice are frequently small or
unlabeled, restricting the full potential of deep learning methods. Here, we
introduce REMEMBER -- Retrieval-based Explainable Multimodal Evidence-guided
Modeling for Brain Evaluation and Reasoning -- a new machine learning framework
that facilitates zero- and few-shot Alzheimer's diagnosis using brain MRI scans
through a reference-based reasoning process. Specifically, REMEMBER first
trains a contrastively aligned vision-text model using expert-annotated
reference data and extends pseudo-text modalities that encode abnormality
types, diagnosis labels, and composite clinical descriptions. Then, at
inference time, REMEMBER retrieves similar, human-validated cases from a
curated dataset and integrates their contextual information through a dedicated
evidence encoding module and attention-based inference head. Such an
evidence-guided design enables REMEMBER to imitate real-world clinical
decision-making process by grounding predictions in retrieved imaging and
textual context. Specifically, REMEMBER outputs diagnostic predictions
alongside an interpretable report, including reference images and explanations
aligned with clinical workflows. Experimental results demonstrate that REMEMBER
achieves robust zero- and few-shot performance and offers a powerful and
explainable framework to neuroimaging-based diagnosis in the real world,
especially under limited data.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 22:06:15 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Can",
"Duy-Cat",
""
],
[
"Tang",
"Quang-Huy",
""
],
[
"Ha",
"Huong",
""
],
[
"Nguyen",
"Binh T.",
""
],
[
"Chén",
"Oliver Y.",
""
]
] |
2504.09355 | Roberta Mota | Roberta Mota, Ehud Sharlin, Usman Alim | Designing Reality-Based VR Interfaces for Geological Uncertainty | null | null | null | null | cs.HC cs.GR | http://creativecommons.org/licenses/by/4.0/ | Inherent uncertainty in geological data acquisition leads to the generation
of large ensembles of equiprobable 3D reservoir models. Running computationally
costly numerical flow simulations across such a vast solution space is
infeasible. A more suitable approach is to carefully select a small number of
geological models that reasonably capture the overall variability of the
ensemble. Identifying these representative models is a critical task that
enables the oil and gas industry to generate cost-effective production
forecasts. Our work leverages virtual reality (VR) to provide engineers with a
system for conducting geological uncertainty analysis, enabling them to perform
inherently spatial tasks using an associative 3D interaction space. We present
our VR system through the lens of the reality-based interaction paradigm,
designing 3D interfaces that enable familiar physical interactions inspired by
real-world analogies-such as gesture-based operations and view-dependent
lenses. We also report an evaluation conducted with 12 reservoir engineers from
an industry partner. Our findings offer insights into the benefits, pitfalls,
and opportunities for refining our system design. We catalog our results into a
set of design recommendations intended to guide researchers and developers of
immersive interfaces-in reservoir engineering and broader application domains.
| [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 22:07:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Mota",
"Roberta",
""
],
[
"Sharlin",
"Ehud",
""
],
[
"Alim",
"Usman",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.