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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2405.07613 | Yuta Kikuchi | Kazuhiro Seki, Yuta Kikuchi, Tomoya Hayata, Seiji Yunoki | Simulating Floquet scrambling circuits on trapped-ion quantum computers | 30 pages, 17 figures, accepted version | Phys. Rev. Research 7, 023032 (2025) | 10.1103/PhysRevResearch.7.023032 | RIKEN-iTHEMS-Report-24 | quant-ph cond-mat.str-el hep-th | http://creativecommons.org/licenses/by/4.0/ | Complex quantum many-body dynamics spread initially localized quantum
information across the entire system. Information scrambling refers to such a
process, whose simulation is one of the promising applications of quantum
computing. We demonstrate the Hayden-Preskill recovery protocol and the
interferometric protocol for calculating out-of-time-ordered correlators to
study the scrambling property of a one-dimensional kicked-Ising model on
20-qubit trapped-ion quantum processors. The simulated quantum circuits have a
geometrically local structure that exhibits the ballistic growth of
entanglement, resulting in the circuit depth being linear in the number of
qubits for the entire state to be scrambled. We experimentally confirm the
growth of signals in the Hayden-Preskill recovery protocol and the decay of
out-of-time-ordered correlators at late times. As an application of the created
scrambling circuits, we also experimentally demonstrate the calculation of the
microcanonical expectation values of local operators adopting the idea of
thermal pure quantum states. Our experiments are made possible by extensively
utilizing one of the highest-fidelity quantum processors currently available
and, thus, should be considered as a benchmark for the current status of the
most advanced quantum computers.
| [
{
"version": "v1",
"created": "Mon, 13 May 2024 10:22:17 GMT"
},
{
"version": "v2",
"created": "Thu, 6 Jun 2024 00:42:31 GMT"
},
{
"version": "v3",
"created": "Mon, 14 Apr 2025 01:58:02 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Seki",
"Kazuhiro",
""
],
[
"Kikuchi",
"Yuta",
""
],
[
"Hayata",
"Tomoya",
""
],
[
"Yunoki",
"Seiji",
""
]
] |
2405.08825 | Yuichiro Terasaki | Yuichiro Terasaki and Kohei Nakajima | Thermodynamic limit in learning period three | 19 pages, 12 figures | null | null | null | stat.ML cs.LG nlin.AO nlin.CD | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A continuous one-dimensional map with period three includes all periods. This
raises the following question: Can we obtain any types of periodic orbits
solely by learning three data points? In this paper, we report the answer to be
yes. Considering a random neural network in its thermodynamic limit, we first
show that almost all learned periods are unstable, and each network has its own
characteristic attractors (which can even be untrained ones). The latently
acquired dynamics, which are unstable within the trained network, serve as a
foundation for the diversity of characteristic attractors and may even lead to
the emergence of attractors of all periods after learning. When the neural
network interpolation is quadratic, a universal post-learning bifurcation
scenario appears, which is consistent with a topological conjugacy between the
trained network and the classical logistic map. In addition to universality, we
explore specific properties of certain networks, including the singular
behavior of the scale of weight at the infinite limit, the finite-size effects,
and the symmetry in learning period three.
| [
{
"version": "v1",
"created": "Sun, 12 May 2024 17:57:25 GMT"
},
{
"version": "v2",
"created": "Thu, 12 Sep 2024 09:22:38 GMT"
},
{
"version": "v3",
"created": "Tue, 28 Jan 2025 12:36:20 GMT"
},
{
"version": "v4",
"created": "Mon, 14 Apr 2025 11:44:36 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Terasaki",
"Yuichiro",
""
],
[
"Nakajima",
"Kohei",
""
]
] |
2405.10952 | Gabriel Moreira | Gabriel Moreira, Manuel Marques, Jo\~ao Paulo Costeira, Alexander
Hauptmann | VICAN: Very Efficient Calibration Algorithm for Large Camera Networks | To appear at the IEEE International Conference on Robotics and
Automation (ICRA), 2024 | null | 10.1109/ICRA57147.2024.10611245 | null | cs.CV cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The precise estimation of camera poses within large camera networks is a
foundational problem in computer vision and robotics, with broad applications
spanning autonomous navigation, surveillance, and augmented reality. In this
paper, we introduce a novel methodology that extends state-of-the-art Pose
Graph Optimization (PGO) techniques. Departing from the conventional PGO
paradigm, which primarily relies on camera-camera edges, our approach centers
on the introduction of a dynamic element - any rigid object free to move in the
scene - whose pose can be reliably inferred from a single image. Specifically,
we consider the bipartite graph encompassing cameras, object poses evolving
dynamically, and camera-object relative transformations at each time step. This
shift not only offers a solution to the challenges encountered in directly
estimating relative poses between cameras, particularly in adverse
environments, but also leverages the inclusion of numerous object poses to
ameliorate and integrate errors, resulting in accurate camera pose estimates.
Though our framework retains compatibility with traditional PGO solvers, its
efficacy benefits from a custom-tailored optimization scheme. To this end, we
introduce an iterative primal-dual algorithm, capable of handling large graphs.
Empirical benchmarks, conducted on a new dataset of simulated indoor
environments, substantiate the efficacy and efficiency of our approach.
| [
{
"version": "v1",
"created": "Mon, 25 Mar 2024 17:47:03 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Moreira",
"Gabriel",
""
],
[
"Marques",
"Manuel",
""
],
[
"Costeira",
"João Paulo",
""
],
[
"Hauptmann",
"Alexander",
""
]
] |
2405.11166 | Takaaki Aoki | Takaaki Aoki, Kohei Nagamachi, Tetsuya Shimane | Learning about the liveability of cities from young migrants using the
combinatiorial Hodge theory approach | null | null | null | null | physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Migration involves making a significant decision to leave one place and
settle in another, entailing substantial career and lifestyle changes.
Migration flows are then the collection of individuals' comparative evaluations
of origin-destination pairs, implicitly revealing people's preferences about
where to live as people ``vote with their feet'' (Tiebout, 1956). However, is
it possible to derive a consistent measure of the liveability of cities from
these flows? We propose a combinatorial-Hodge-theory approach: the empirical
liveability of cities is evaluated by a potential governing unbalanced, acyclic
migrations between cities. As a case study, we measure the liveability of
municipalities in Japan for specific populations such as families with small
children and women of reproductive age in a population-decline society. Using
these potentials as dependent variables, we perform a regression analysis to
identify the factors relevant to liveability. We also derive analytical
expressions that allow us to interpret as potentials the standards of living or
utilities, estimated in the economics literature (Douglas & Wall, 1993;
Douglas, 1997; Douglas & Wall, 2000; Wall, 2001; Nakajima & Tabuchi, 2011). The
proposed method extracts a consistent metric of interval scale from the
non-transitive, pairwise comparison between locations and provides substantial
statistics for urban planning by policymakers.
| [
{
"version": "v1",
"created": "Sat, 18 May 2024 03:50:16 GMT"
},
{
"version": "v2",
"created": "Tue, 8 Oct 2024 00:18:49 GMT"
},
{
"version": "v3",
"created": "Sun, 13 Apr 2025 04:00:01 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Aoki",
"Takaaki",
""
],
[
"Nagamachi",
"Kohei",
""
],
[
"Shimane",
"Tetsuya",
""
]
] |
2405.11260 | Guodong Jiang | Guodong Jiang and Yafis Barlas | Geometric superfluid weight of composite bands in multiorbital
superconductors | null | null | 10.1103/PhysRevB.109.214518 | null | cond-mat.supr-con | http://creativecommons.org/licenses/by/4.0/ | The superfluid weight of an isolated flat band in multi-orbital
superconductors contains contributions from the band's quantum metric and a
lattice geometric term that depends on the orbital positions in the lattice.
Since the superfluid weight is a measure of the superconductor's energy
fluctuation, it is independent of the lattice geometry, leading to the minimal
quantum metric of a band [Phys. Rev. B 106, 014518 (2022)]. Here, a
perturbation approach is developed to study the superfluid weight and its
lattice geometric dependence for composite bands. When all orbitals exhibit
uniform pairing, the quantum geometric term contains each band's contribution
and an inter-band contribution between every pair of bands in the composite.
Based on a band representation analysis, they provide a topological lower bound
for the superfluid weight of an isolated composite of flat bands. Using this
perturbation approach, an analytical expression of the lattice geometric
contribution is obtained. It is expressed in terms of Bloch functions,
providing a convenient formula to calculate the superfluid weight for
multi-orbital superconductors.
| [
{
"version": "v1",
"created": "Sat, 18 May 2024 11:16:18 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 14:56:53 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Jiang",
"Guodong",
""
],
[
"Barlas",
"Yafis",
""
]
] |
2405.12405 | Sara Haravifard | Lalit Yadav, Afonso Rufino, Rabindranath Bag, Matthew Ennis, Jan
Alexander Koziol, Clarina dela Cruz, Alexander I. Kolesnikov, V. Ovidiu
Garlea, Keith M. Taddei, David Graf, Kai Phillip Schmidt, Fr\'ed\'eric Mila,
and Sara Haravifard | Observation of Unprecedented Fractional Magnetization Plateaus in a New
Shastry-Sutherland Ising Compound | null | null | null | null | cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Geometrically frustrated magnetic systems, such as those based on the
Shastry-Sutherland lattice (SSL), offer a rich playground for exploring
unconventional magnetic states. The delicate balance between competing
interactions in these systems leads to the emergence of novel phases. We
present the characterization of Er2Be2GeO7, an SSL compound with Er3+ ions
forming orthogonal dimers separated by non-magnetic layers whose structure is
invariant under the P-421m space group. Neutron scattering reveals an
antiferromagnetic dimer structure at zero field, typical of Ising spins on that
lattice and consistent with the anisotropic magnetization observed. However,
magnetization measurements exhibit fractional plateaus at 1/4 and 1/2 of
saturation, in contrast to the expected 1/3 plateau of the SSL Ising model. By
comparing the energy of candidate states with ground-state lower bounds we show
that this behavior requires spatially anisotropic interactions, leading to an
anisotropic Shastry-Sutherland Ising Model (ASSLIM) symmetric under the Cmm2
space group. This anisotropy is consistent with the small orthorhombic
distortion observed with single-crystal neutron diffraction. The other
properties, including thermodynamics, which have been investigated
theoretically using tensor networks, point to small residual interactions,
potentially due to further couplings and quantum fluctuations. This study
highlights Er2Ge2GeO7 as a promising platform for investigating exotic magnetic
phenomena.
| [
{
"version": "v1",
"created": "Mon, 20 May 2024 23:02:40 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 21:00:04 GMT"
},
{
"version": "v3",
"created": "Sat, 12 Apr 2025 13:57:28 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Yadav",
"Lalit",
""
],
[
"Rufino",
"Afonso",
""
],
[
"Bag",
"Rabindranath",
""
],
[
"Ennis",
"Matthew",
""
],
[
"Koziol",
"Jan Alexander",
""
],
[
"Cruz",
"Clarina dela",
""
],
[
"Kolesnikov",
"Alexander I.",
""
],
[
"Garlea",
"V. Ovidiu",
""
],
[
"Taddei",
"Keith M.",
""
],
[
"Graf",
"David",
""
],
[
"Schmidt",
"Kai Phillip",
""
],
[
"Mila",
"Frédéric",
""
],
[
"Haravifard",
"Sara",
""
]
] |
2405.13506 | Aitor R. Gomez | Aitor R. Gomez, Manuela L. Bujorianu, Rafal Wisniewski | Large Deviations in Safety-Critical Systems with Probabilistic Initial
Conditions | null | null | null | null | math.OC cond-mat.stat-mech | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We often rely on probabilistic measures--e.g. event probability or expected
time--to characterize systems safety. However, determining these quantities for
extremely low-probability events is generally challenging, as standard safety
methods usually struggle due to conservativeness, high-dimension scalability,
tractability or numerical limitations. We address these issues by leveraging
rigorous approximations grounded in the principles of Large Deviations theory.
By assuming deterministic initial conditions, Large Deviations identifies a
single dominant path as the most significant contributor to the rare-event
probability: the instanton. We extend this result to incorporate stochastic
uncertainty in the initial states, which is a common assumption in many
applications. To that end, we determine an expression for the probability
density of the initial states, conditioned on the instanton--the most dominant
path hitting the unsafe region--being observed. This expression gives access to
the most probable initial conditions, as well as the most probable hitting time
and path deviations, leading to an unsafe rare event. We demonstrate its
effectiveness by solving a high-dimensional and non-linear problem: a space
collision.
| [
{
"version": "v1",
"created": "Wed, 22 May 2024 10:06:13 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 17:22:19 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Gomez",
"Aitor R.",
""
],
[
"Bujorianu",
"Manuela L.",
""
],
[
"Wisniewski",
"Rafal",
""
]
] |
2405.13637 | Radu Tudor Ionescu | Florinel-Alin Croitoru, Vlad Hondru, Radu Tudor Ionescu, Nicu Sebe,
Mubarak Shah | Curriculum Direct Preference Optimization for Diffusion and Consistency
Models | Accepted at CVPR 2025 | null | null | null | cs.CV cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Direct Preference Optimization (DPO) has been proposed as an effective and
efficient alternative to reinforcement learning from human feedback (RLHF). In
this paper, we propose a novel and enhanced version of DPO based on curriculum
learning for text-to-image generation. Our method is divided into two training
stages. First, a ranking of the examples generated for each prompt is obtained
by employing a reward model. Then, increasingly difficult pairs of examples are
sampled and provided to a text-to-image generative (diffusion or consistency)
model. Generated samples that are far apart in the ranking are considered to
form easy pairs, while those that are close in the ranking form hard pairs. In
other words, we use the rank difference between samples as a measure of
difficulty. The sampled pairs are split into batches according to their
difficulty levels, which are gradually used to train the generative model. Our
approach, Curriculum DPO, is compared against state-of-the-art fine-tuning
approaches on nine benchmarks, outperforming the competing methods in terms of
text alignment, aesthetics and human preference. Our code is available at
https://github.com/CroitoruAlin/Curriculum-DPO.
| [
{
"version": "v1",
"created": "Wed, 22 May 2024 13:36:48 GMT"
},
{
"version": "v2",
"created": "Fri, 24 May 2024 13:14:40 GMT"
},
{
"version": "v3",
"created": "Tue, 11 Mar 2025 16:44:48 GMT"
},
{
"version": "v4",
"created": "Tue, 18 Mar 2025 13:03:48 GMT"
},
{
"version": "v5",
"created": "Sat, 12 Apr 2025 11:26:34 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Croitoru",
"Florinel-Alin",
""
],
[
"Hondru",
"Vlad",
""
],
[
"Ionescu",
"Radu Tudor",
""
],
[
"Sebe",
"Nicu",
""
],
[
"Shah",
"Mubarak",
""
]
] |
2405.13685 | Divya Kothandaraman | Divya Kothandaraman, Ming Lin, Dinesh Manocha | Financial Models in Generative Art: Black-Scholes-Inspired Concept
Blending in Text-to-Image Diffusion | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | We introduce a novel approach for concept blending in pretrained
text-to-image diffusion models, aiming to generate images at the intersection
of multiple text prompts. At each time step during diffusion denoising, our
algorithm forecasts predictions w.r.t. the generated image and makes informed
text conditioning decisions. Central to our method is the unique analogy
between diffusion models, which are rooted in non-equilibrium thermodynamics,
and the Black-Scholes model for financial option pricing. By drawing parallels
between key variables in both domains, we derive a robust algorithm for concept
blending that capitalizes on the Markovian dynamics of the Black-Scholes
framework. Our text-based concept blending algorithm is data-efficient, meaning
it does not need additional training. Furthermore, it operates without human
intervention or hyperparameter tuning. We highlight the benefits of our
approach by comparing it qualitatively and quantitatively to other text based
concept blending techniques, including linear interpolation, alternating
prompts, step-wise prompt switching, and CLIP-guided prompt selection across
various scenarios such as single object per text prompt, multiple objects per
text prompt and objects against backgrounds. Our work shows that financially
inspired techniques can enhance text-to-image concept blending in generative
AI, paving the way for broader innovation. Code is available at
https://github.com/divyakraman/BlackScholesDiffusion2024.
| [
{
"version": "v1",
"created": "Wed, 22 May 2024 14:25:57 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 21:43:05 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Kothandaraman",
"Divya",
""
],
[
"Lin",
"Ming",
""
],
[
"Manocha",
"Dinesh",
""
]
] |
2405.13721 | Zhiwei Bai | Zhiwei Bai, Jiajie Zhao, Yaoyu Zhang | Connectivity Shapes Implicit Regularization in Matrix Factorization
Models for Matrix Completion | 34 pages | 38th Conference on Neural Information Processing Systems (NeurIPS
2024) | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Matrix factorization models have been extensively studied as a valuable
test-bed for understanding the implicit biases of overparameterized models.
Although both low nuclear norm and low rank regularization have been studied
for these models, a unified understanding of when, how, and why they achieve
different implicit regularization effects remains elusive. In this work, we
systematically investigate the implicit regularization of matrix factorization
for solving matrix completion problems. We empirically discover that the
connectivity of observed data plays a crucial role in the implicit bias, with a
transition from low nuclear norm to low rank as data shifts from disconnected
to connected with increased observations. We identify a hierarchy of intrinsic
invariant manifolds in the loss landscape that guide the training trajectory to
evolve from low-rank to higher-rank solutions. Based on this finding, we
theoretically characterize the training trajectory as following the
hierarchical invariant manifold traversal process, generalizing the
characterization of Li et al. (2020) to include the disconnected case.
Furthermore, we establish conditions that guarantee minimum nuclear norm,
closely aligning with our experimental findings, and we provide a dynamics
characterization condition for ensuring minimum rank. Our work reveals the
intricate interplay between data connectivity, training dynamics, and implicit
regularization in matrix factorization models.
| [
{
"version": "v1",
"created": "Wed, 22 May 2024 15:12:14 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 08:49:39 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Bai",
"Zhiwei",
""
],
[
"Zhao",
"Jiajie",
""
],
[
"Zhang",
"Yaoyu",
""
]
] |
2405.13929 | Aleksandr Nikolich | Aleksandr Nikolich, Konstantin Korolev, Sergei Bratchikov, Igor
Kiselev, Artem Shelmanov | Vikhr: The Family of Open-Source Instruction-Tuned Large Language Models
for Russian | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | There has been a surge in the development of various Large Language Models
(LLMs). However, text generation for languages other than English often faces
significant challenges, including poor generation quality and reduced
computational performance due to the disproportionate representation of tokens
in the model's vocabulary. In this work, we address these issues by developing
a pipeline for the adaptation of English-oriented pre-trained models to other
languages and constructing efficient bilingual LLMs. Using this pipeline, we
construct Vikhr, a series of bilingual open-source instruction-following LLMs
designed specifically for the Russian language. ``Vikhr'' refers to the name of
the Mistral LLM series and means a ``strong gust of wind.'' Unlike previous
Russian-language models that typically rely on LoRA adapters on top of
English-oriented models, sacrificing performance for lower training costs,
Vikhr features an adapted tokenizer vocabulary and undergoes the continued
pre-training and instruction tuning of all weights. This not only enhances the
model's performance but also significantly improves its computational and
contextual efficiency. We also expanded the instruction datasets and corpora
for continued pre-training. The model weights, instruction sets, and code are
publicly available.
| [
{
"version": "v1",
"created": "Wed, 22 May 2024 18:58:58 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Jun 2024 17:32:23 GMT"
},
{
"version": "v3",
"created": "Sat, 26 Oct 2024 08:47:36 GMT"
},
{
"version": "v4",
"created": "Wed, 13 Nov 2024 10:57:21 GMT"
},
{
"version": "v5",
"created": "Mon, 24 Feb 2025 13:24:20 GMT"
},
{
"version": "v6",
"created": "Mon, 14 Apr 2025 12:23:27 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Nikolich",
"Aleksandr",
""
],
[
"Korolev",
"Konstantin",
""
],
[
"Bratchikov",
"Sergei",
""
],
[
"Kiselev",
"Igor",
""
],
[
"Shelmanov",
"Artem",
""
]
] |
2405.14139 | Xinhao Fan | Xinhao Fan, Shreesh P Mysore | Contribute to balance, wire in accordance: Emergence of backpropagation
from a simple, bio-plausible neuroplasticity rule | null | null | null | null | q-bio.NC cs.LG cs.NE | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Backpropagation (BP) has been pivotal in advancing machine learning and
remains essential in computational applications and comparative studies of
biological and artificial neural networks. Despite its widespread use, the
implementation of BP in the brain remains elusive, and its biological
plausibility is often questioned due to inherent issues such as the need for
symmetry of weights between forward and backward connections, and the
requirement of distinct forward and backward phases of computation. Here, we
introduce a novel neuroplasticity rule that offers a potential mechanism for
implementing BP in the brain. Similar in general form to the classical Hebbian
rule, this rule is based on the core principles of maintaining the balance of
excitatory and inhibitory inputs as well as on retrograde signaling, and
operates over three progressively slower timescales: neural firing, retrograde
signaling, and neural plasticity. We hypothesize that each neuron possesses an
internal state, termed credit, in addition to its firing rate. After achieving
equilibrium in firing rates, neurons receive credits based on their
contribution to the E-I balance of postsynaptic neurons through retrograde
signaling. As the network's credit distribution stabilizes, connections from
those presynaptic neurons are strengthened that significantly contribute to the
balance of postsynaptic neurons. We demonstrate mathematically that our
learning rule precisely replicates BP in layered neural networks without any
approximations. Simulations on artificial neural networks reveal that this rule
induces varying community structures in networks, depending on the learning
rate. This simple theoretical framework presents a biologically plausible
implementation of BP, with testable assumptions and predictions that may be
evaluated through biological experiments.
| [
{
"version": "v1",
"created": "Thu, 23 May 2024 03:28:52 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 02:34:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Fan",
"Xinhao",
""
],
[
"Mysore",
"Shreesh P",
""
]
] |
2405.15192 | Lingling Chen | Lingling Chen, Mikyoung Jun, Scott J. Cook | Addressing Duplicated Data in Spatial Point Patterns | null | null | null | null | stat.ME | http://creativecommons.org/licenses/by/4.0/ | Spatial point process models are widely applied to point pattern data from
various applications in the social and environmental sciences. However, a
serious hurdle in fitting point process models is the presence of duplicated
points, wherein multiple observations share identical spatial coordinates. This
often occurs because of decisions made in the geo-coding process, such as
assigning representative locations (e.g., aggregate-level centroids) to
observations when data producers lack exact location information. Because
spatial point process models like the Log-Gaussian Cox Process (LGCP) assume
unique locations, researchers often employ ad hoc solutions (e.g., removing
duplicates or jittering) to address duplicated data before analysis. As an
alternative, this study proposes a Modified Minimum Contrast (MMC) method that
adapts the inference procedure to account for the effect of duplicates in
estimation, without needing to alter the data. The proposed MMC method is
applied to LGCP models, focusing on the inference of second-order intensity
parameters, which govern the clustering structure of point patterns. Under a
variety of simulated conditions, our results demonstrate the advantages of the
proposed MMC method compared to existing ad hoc solutions. We then apply the
MMC methods to a real-data application of conflict events in Afghanistan
(2008-2009).
| [
{
"version": "v1",
"created": "Fri, 24 May 2024 03:52:44 GMT"
},
{
"version": "v2",
"created": "Fri, 30 Aug 2024 03:04:25 GMT"
},
{
"version": "v3",
"created": "Fri, 11 Apr 2025 20:42:20 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Chen",
"Lingling",
""
],
[
"Jun",
"Mikyoung",
""
],
[
"Cook",
"Scott J.",
""
]
] |
2405.15391 | Fabio Scarabotti | Fabio Scarabotti | Representation theory of the group of automorphisms of a finite rooted
tree | null | null | null | null | math.RT | http://creativecommons.org/licenses/by/4.0/ | We construct the ordinary irreducible representations of the group of
automorphisms of a finite rooted tree and we get a natural parametrization of
them. To achieve this goals, we introduce and study the combinatorics of tree
compositions, a natural generalization of set compositions but with new
features and more complexity. These combinatorial structures lead to a family
of permutation representations which have the same parametrization of the
irreducible representations. Our trees are not necessarily spherically
homogeneous and our approach is coordinate free.
| [
{
"version": "v1",
"created": "Fri, 24 May 2024 09:43:49 GMT"
},
{
"version": "v2",
"created": "Sun, 23 Jun 2024 00:29:05 GMT"
},
{
"version": "v3",
"created": "Fri, 11 Apr 2025 12:38:04 GMT"
},
{
"version": "v4",
"created": "Mon, 14 Apr 2025 11:12:27 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Scarabotti",
"Fabio",
""
]
] |
2405.15540 | Jacob Bamberger | Jacob Bamberger, Federico Barbero, Xiaowen Dong, Michael M. Bronstein | Bundle Neural Networks for message diffusion on graphs | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | The dominant paradigm for learning on graph-structured data is message
passing. Despite being a strong inductive bias, the local message passing
mechanism suffers from pathological issues such as over-smoothing,
over-squashing, and limited node-level expressivity. To address these
limitations we propose Bundle Neural Networks (BuNN), a new type of GNN that
operates via message diffusion over flat vector bundles - structures analogous
to connections on Riemannian manifolds that augment the graph by assigning to
each node a vector space and an orthogonal map. A BuNN layer evolves the
features according to a diffusion-type partial differential equation. When
discretized, BuNNs are a special case of Sheaf Neural Networks (SNNs), a
recently proposed MPNN capable of mitigating over-smoothing. The continuous
nature of message diffusion enables BuNNs to operate on larger scales of the
graph and, therefore, to mitigate over-squashing. Finally, we prove that BuNN
can approximate any feature transformation over nodes on any (potentially
infinite) family of graphs given injective positional encodings, resulting in
universal node-level expressivity. We support our theory via synthetic
experiments and showcase the strong empirical performance of BuNNs over a range
of real-world tasks, achieving state-of-the-art results on several standard
benchmarks in transductive and inductive settings.
| [
{
"version": "v1",
"created": "Fri, 24 May 2024 13:28:48 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 07:52:41 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Bamberger",
"Jacob",
""
],
[
"Barbero",
"Federico",
""
],
[
"Dong",
"Xiaowen",
""
],
[
"Bronstein",
"Michael M.",
""
]
] |
2405.16213 | Gabriel Moreira | Gabriel Moreira, Manuel Marques, Jo\~ao Paulo Costeira, Alexander
Hauptmann | Learning Visual-Semantic Subspace Representations | The 28th International Conference on Artificial Intelligence and
Statistics (AISTATS) | null | null | null | cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Learning image representations that capture rich semantic relationships
remains a significant challenge. Existing approaches are either contrastive,
lacking robust theoretical guarantees, or struggle to effectively represent the
partial orders inherent to structured visual-semantic data. In this paper, we
introduce a nuclear norm-based loss function, grounded in the same information
theoretic principles that have proved effective in self-supervised learning. We
present a theoretical characterization of this loss, demonstrating that, in
addition to promoting class orthogonality, it encodes the spectral geometry of
the data within a subspace lattice. This geometric representation allows us to
associate logical propositions with subspaces, ensuring that our learned
representations adhere to a predefined symbolic structure.
| [
{
"version": "v1",
"created": "Sat, 25 May 2024 12:51:38 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 17:08:18 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Moreira",
"Gabriel",
""
],
[
"Marques",
"Manuel",
""
],
[
"Costeira",
"João Paulo",
""
],
[
"Hauptmann",
"Alexander",
""
]
] |
2405.16519 | Tal Amir | Tal Amir, Nadav Dym | Fourier Sliced-Wasserstein Embedding for Multisets and Measures | ICLR 2025 camera-ready | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | We present the Fourier Sliced-Wasserstein (FSW) embedding - a novel method to
embed multisets and measures over R^d into Euclidean space.
Our proposed embedding approximately preserves the sliced Wasserstein
distance on distributions, thereby yielding geometrically meaningful
representations that better capture the structure of the input. Moreover, it is
injective on measures and bi-Lipschitz on multisets - a significant advantage
over prevalent methods based on sum- or max-pooling, which are provably not
bi-Lipschitz, and, in many cases, not even injective. The required output
dimension for these guarantees is near-optimal: roughly 2Nd, where N is the
maximal input multiset size.
Furthermore, we prove that it is impossible to embed distributions over R^d
into Euclidean space in a bi-Lipschitz manner. Thus, the metric properties of
our embedding are, in a sense, the best possible.
Through numerical experiments, we demonstrate that our method yields superior
multiset representations that improve performance in practical learning tasks.
Specifically, we show that (a) a simple combination of the FSW embedding with
an MLP achieves state-of-the-art performance in learning the (non-sliced)
Wasserstein distance; and (b) replacing max-pooling with the FSW embedding
makes PointNet significantly more robust to parameter reduction, with only
minor performance degradation even after a 40-fold reduction.
| [
{
"version": "v1",
"created": "Sun, 26 May 2024 11:04:41 GMT"
},
{
"version": "v2",
"created": "Sun, 29 Sep 2024 22:36:41 GMT"
},
{
"version": "v3",
"created": "Mon, 14 Apr 2025 13:33:41 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Amir",
"Tal",
""
],
[
"Dym",
"Nadav",
""
]
] |
2405.16555 | Zhaozhi Wang | Zhaozhi Wang, Yue Liu, Yunjie Tian, Yunfan Liu, Yaowei Wang, Qixiang
Ye | Building Vision Models upon Heat Conduction | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Visual representation models leveraging attention mechanisms are challenged
by significant computational overhead, particularly when pursuing large
receptive fields. In this study, we aim to mitigate this challenge by
introducing the Heat Conduction Operator (HCO) built upon the physical heat
conduction principle. HCO conceptualizes image patches as heat sources and
models their correlations through adaptive thermal energy diffusion, enabling
robust visual representations. HCO enjoys a computational complexity of
O(N^1.5), as it can be implemented using discrete cosine transformation (DCT)
operations. HCO is plug-and-play, combining with deep learning backbones
produces visual representation models (termed vHeat) with global receptive
fields. Experiments across vision tasks demonstrate that, beyond the stronger
performance, vHeat achieves up to a 3x throughput, 80% less GPU memory
allocation, and 35% fewer computational FLOPs compared to the Swin-Transformer.
Code is available at https://github.com/MzeroMiko/vHeat.
| [
{
"version": "v1",
"created": "Sun, 26 May 2024 12:58:04 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 10:44:13 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Wang",
"Zhaozhi",
""
],
[
"Liu",
"Yue",
""
],
[
"Tian",
"Yunjie",
""
],
[
"Liu",
"Yunfan",
""
],
[
"Wang",
"Yaowei",
""
],
[
"Ye",
"Qixiang",
""
]
] |
2405.18869 | Vid Han\v{z}el | Vid Han\v{z}el, Bla\v{z} Bertalani\v{c}, Carolina Fortuna | Towards Data-Driven Electricity Management: Multi-Region Harmonized Data
and Knowledge Graph | Submitted to: Scientific Data | null | 10.1038/s41597-024-04310-z | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Due to growing population and technological advances, global electricity
consumption, and consequently also CO2 emissions are increasing. The
residential sector makes up 25% of global electricity consumption and has great
potential to increase efficiency and reduce CO2 footprint without sacrificing
comfort. However, a lack of uniform consumption data at the household level
spanning multiple regions hinders large-scale studies and robust multi-region
model development. This paper introduces a multi-region dataset compiled from
publicly available sources and presented in a uniform format. This data enables
machine learning tasks such as disaggregation, demand forecasting, appliance
ON/OFF classification, etc. Furthermore, we develop an RDF knowledge graph that
characterizes the electricity consumption of the households and contextualizes
it with household related properties enabling semantic queries and
interoperability with other open knowledge bases like Wikidata and DBpedia.
This structured data can be utilized to inform various stakeholders towards
data-driven policy and business development.
| [
{
"version": "v1",
"created": "Wed, 29 May 2024 08:30:34 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hanžel",
"Vid",
""
],
[
"Bertalanič",
"Blaž",
""
],
[
"Fortuna",
"Carolina",
""
]
] |
2405.20171 | Garry Goldstein | Garry Goldstein | Projected Augmented Waves (PAW): extended resolution of unity method | Comments welcome, V2: Many corrections, V3: Minor typos corrected,
V4:some minor typos | null | null | null | cond-mat.other cond-mat.mtrl-sci physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | The Projected Augmented Waves (PAW) method is based on a linear
transformation between the pseudo wavefunctions and the all electron
wavefunctions. To obtain high accuracy with this method, it is important that
the local part of the linear transform (inside each atomic sphere) be defined
over a complete basis set (with deviations from completeness leading to
corrections to the total energy not computed within current implementations of
PAW). Here we show how to make this basis much closer to complete without
significant additional computational work and without modifying the
transformation in any significant way thereby making the modifications we
propose easy to implement in current electronic structure codes for PAW.. This
is done by extending the resolution of unity used for the transform to include
more smooth wavefunctions (which have nothing to do with the atomic problem)
and having them linearly transform via the identity.
| [
{
"version": "v1",
"created": "Thu, 30 May 2024 15:43:49 GMT"
},
{
"version": "v2",
"created": "Sun, 7 Jul 2024 17:16:59 GMT"
},
{
"version": "v3",
"created": "Sun, 16 Feb 2025 23:37:20 GMT"
},
{
"version": "v4",
"created": "Sat, 12 Apr 2025 21:28:21 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Goldstein",
"Garry",
""
]
] |
2405.20179 | Zichao Hu | Zichao Hu, Junyi Jessy Li, Arjun Guha, Joydeep Biswas | Robo-Instruct: Simulator-Augmented Instruction Alignment For Finetuning
Code LLMs | null | null | null | null | cs.CL cs.AI cs.RO | http://creativecommons.org/licenses/by/4.0/ | Code LLMs have shown promising results with converting tasks in natural
language to programs that can be executed by service robots. We are interested
in finetuning small, specialized LLMs for this purpose, but collecting datasets
of task-program pairs specific to each robot is time-consuming and expensive.
While approaches such as SELF-INSTRUCT and EVOL-INSTRUCT are capable of
generating novel tasks given a few examples, they are unable to provide the
corresponding programs that correctly abide by physical-world and
robot-constraints using the provided programming interface. Using a simulator
is a natural potential solution to checking for such constraints, but building
simulation environments that can handle arbitrary tasks and their necessary
objects and locations, is challenging. To address these challenges, we
introduce ROBO-INSTRUCT, which synthesizes task-specific simulation
environments on the fly during program execution, by opportunistically
inferring entity properties and enforcing corresponding constraints based on
how the entities are used in the task program. Additionally, ROBO-INSTRUCT
integrates an LLM-aided post-processing procedure to refine instructions for
better alignment with robot programs. We demonstrate the effectiveness of
ROBO-INSTRUCT across multiple LLMs, showing that our fine-tuned models
outperform all baseline methods and even match or surpass the performance of
several larger and proprietary models.
| [
{
"version": "v1",
"created": "Thu, 30 May 2024 15:47:54 GMT"
},
{
"version": "v2",
"created": "Sat, 5 Oct 2024 23:27:10 GMT"
},
{
"version": "v3",
"created": "Fri, 11 Apr 2025 19:55:48 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hu",
"Zichao",
""
],
[
"Li",
"Junyi Jessy",
""
],
[
"Guha",
"Arjun",
""
],
[
"Biswas",
"Joydeep",
""
]
] |
2406.00104 | Samuel Duffield | Samuel Duffield, Kaelan Donatella, Johnathan Chiu, Phoebe Klett,
Daniel Simpson | Scalable Bayesian Learning with posteriors | null | Published as a conference paper at ICLR 2025 | null | null | cs.LG stat.ML | http://creativecommons.org/licenses/by/4.0/ | Although theoretically compelling, Bayesian learning with modern machine
learning models is computationally challenging since it requires approximating
a high dimensional posterior distribution. In this work, we (i) introduce
posteriors, an easily extensible PyTorch library hosting general-purpose
implementations making Bayesian learning accessible and scalable to large data
and parameter regimes; (ii) present a tempered framing of stochastic gradient
Markov chain Monte Carlo, as implemented in posteriors, that transitions
seamlessly into optimization and unveils a minor modification to deep ensembles
to ensure they are asymptotically unbiased for the Bayesian posterior, and
(iii) demonstrate and compare the utility of Bayesian approximations through
experiments including an investigation into the cold posterior effect and
applications with large language models.
| [
{
"version": "v1",
"created": "Fri, 31 May 2024 18:00:12 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 16:36:07 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Duffield",
"Samuel",
""
],
[
"Donatella",
"Kaelan",
""
],
[
"Chiu",
"Johnathan",
""
],
[
"Klett",
"Phoebe",
""
],
[
"Simpson",
"Daniel",
""
]
] |
2406.01320 | Yumiharu Nakano | Yumiharu Nakano | Convergence of the denoising diffusion probabilistic models for general
noise schedules | Substantial changes have been made in this version | null | null | null | math.PR stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work presents a theoretical analysis of the original formulation of
denoising diffusion probabilistic models (DDPMs), introduced by Ho, Jain, and
Abbeel in Advances in Neural Information Processing Systems, 33 (2020), pp.
6840-6851. An explicit upper bound is derived for the total variation distance
between the distribution of the discrete-time DDPM sampling algorithm and a
target data distribution, under general noise schedule parameters. The analysis
assumes certain technical conditions on the data distribution and a linear
growth condition on the noise estimation function. The sampling sequence
emerges as an exponential integrator-type approximation of a reverse-time
stochastic differential equation (SDE) over a finite time interval.
Schr\"odinger's problem provides a tool for estimating the distributional error
in reverse time, which connects the reverse-time error with its forward-time
counterpart. The score function in DDPMs appears as an adapted solution of a
forward-backward SDE, providing a foundation for analyzing the
time-discretization error associated with the reverse-time SDE.
| [
{
"version": "v1",
"created": "Mon, 3 Jun 2024 13:38:18 GMT"
},
{
"version": "v2",
"created": "Fri, 6 Sep 2024 01:30:50 GMT"
},
{
"version": "v3",
"created": "Tue, 5 Nov 2024 06:11:25 GMT"
},
{
"version": "v4",
"created": "Tue, 14 Jan 2025 08:11:38 GMT"
},
{
"version": "v5",
"created": "Mon, 14 Apr 2025 03:36:15 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Nakano",
"Yumiharu",
""
]
] |
2406.01660 | Mohammad Gheshlaghi Azar | Eugene Choi, Arash Ahmadian, Matthieu Geist, Oilvier Pietquin,
Mohammad Gheshlaghi Azar | Self-Improving Robust Preference Optimization | null | null | null | null | cs.LG cs.AI stat.ML | http://creativecommons.org/licenses/by/4.0/ | Online and offline RLHF methods, such as PPO and DPO, have been highly
successful in aligning AI with human preferences. Despite their success,
however, these methods suffer from fundamental limitations: (a) Models trained
with RLHF can learn from mistakes or negative examples through RL mechanism or
contrastive loss during training. However, at inference time, they lack an
innate self-improvement mechanism for error corrections. (b) The optimal
solution of existing methods is highly task-dependent, making it difficult for
them to generalize to new tasks. To address these challenges, we propose
Self-Improving Robust Preference Optimization (SRPO), a practical and
mathematically principled offline RLHF framework. The key idea behind SRPO is
to cast the problem of learning from human preferences as a self-improvement
process, mathematically formulated as a min-max objective that jointly
optimizes a self-improvement policy and a generative policy in an adversarial
fashion. Crucially, the solution for this optimization problem is independent
of the training task, which makes it robust to its changes. We then show that
this objective can be reformulated as a non-adversarial offline loss, which can
be efficiently optimized using standard supervised learning techniques at
scale. To demonstrate SRPO's effectiveness, we evaluate it using AI Win-Rate
(WR) against human (GOLD) completions. When tested on the XSum dataset, SRPO
outperforms DPO by a margin of 15% after 5 self revisions, achieving an
impressive 90% WR. Moreover, on the challenging Arena-Hard prompts, SRPO
outperforms both DPO and IPO (by 4% without revision and 6% after a single
revision), reaching a 56% WR against against Llama-3.1-8B-Instruct.
| [
{
"version": "v1",
"created": "Mon, 3 Jun 2024 17:53:25 GMT"
},
{
"version": "v2",
"created": "Wed, 5 Jun 2024 01:25:34 GMT"
},
{
"version": "v3",
"created": "Fri, 7 Jun 2024 17:25:12 GMT"
},
{
"version": "v4",
"created": "Fri, 11 Apr 2025 23:24:37 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Choi",
"Eugene",
""
],
[
"Ahmadian",
"Arash",
""
],
[
"Geist",
"Matthieu",
""
],
[
"Pietquin",
"Oilvier",
""
],
[
"Azar",
"Mohammad Gheshlaghi",
""
]
] |
2406.01708 | Ihsen Alouani | Mahmoud Ghorbel, Halima Bouzidi, Ioan Marius Bilasco, Ihsen Alouani | SnatchML: Hijacking ML models without Training Access | 17 pages, 14 figures, 7 tables | null | null | null | cs.CR cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Model hijacking can cause significant accountability and security risks since
the owner of a hijacked model can be framed for having their model offer
illegal or unethical services. Prior works consider model hijacking as a
training time attack, whereby an adversary requires full access to the ML model
training. In this paper, we consider a stronger threat model for an
inference-time hijacking attack, where the adversary has no access to the
training phase of the victim model. Our intuition is that ML models, which are
typically over-parameterized, might have the capacity to (unintentionally)
learn more than the intended task they are trained for. We propose SnatchML, a
new training-free model hijacking attack, that leverages the extra capacity
learnt by the victim model to infer different tasks that can be semantically
related or unrelated to the original one. Our results on models deployed on AWS
Sagemaker showed that SnatchML can deliver high accuracy on hijacking tasks.
Interestingly, while all previous approaches are limited by the number of
classes in the benign task, SnatchML can hijack models for tasks that contain
more classes than the original. We explore different methods to mitigate this
risk; We propose meta-unlearning, which is designed to help the model unlearn a
potentially malicious task while training for the original task. We also
provide insights on over-parametrization as a possible inherent factor that
facilitates model hijacking, and accordingly, we propose a compression-based
countermeasure to counteract this attack. We believe this work offers a
previously overlooked perspective on model hijacking attacks, presenting a
stronger threat model and higher applicability in real-world contexts.
| [
{
"version": "v1",
"created": "Mon, 3 Jun 2024 18:04:37 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 09:56:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ghorbel",
"Mahmoud",
""
],
[
"Bouzidi",
"Halima",
""
],
[
"Bilasco",
"Ioan Marius",
""
],
[
"Alouani",
"Ihsen",
""
]
] |
2406.01867 | Kengo Uchida | Kengo Uchida, Takashi Shibuya, Yuhta Takida, Naoki Murata, Julian
Tanke, Shusuke Takahashi, Yuki Mitsufuji | MoLA: Motion Generation and Editing with Latent Diffusion Enhanced by
Adversarial Training | CVPR 2025 HuMoGen Workshop | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | In text-to-motion generation, controllability as well as generation quality
and speed has become increasingly critical. The controllability challenges
include generating a motion of a length that matches the given textual
description and editing the generated motions according to control signals,
such as the start-end positions and the pelvis trajectory. In this paper, we
propose MoLA, which provides fast, high-quality, variable-length motion
generation and can also deal with multiple editing tasks in a single framework.
Our approach revisits the motion representation used as inputs and outputs in
the model, incorporating an activation variable to enable variable-length
motion generation. Additionally, we integrate a variational autoencoder and a
latent diffusion model, further enhanced through adversarial training, to
achieve high-quality and fast generation. Moreover, we apply a training-free
guided generation framework to achieve various editing tasks with motion
control inputs. We quantitatively show the effectiveness of adversarial
learning in text-to-motion generation, and demonstrate the applicability of our
editing framework to multiple editing tasks in the motion domain.
| [
{
"version": "v1",
"created": "Tue, 4 Jun 2024 00:38:44 GMT"
},
{
"version": "v2",
"created": "Thu, 18 Jul 2024 07:45:54 GMT"
},
{
"version": "v3",
"created": "Mon, 17 Feb 2025 09:00:41 GMT"
},
{
"version": "v4",
"created": "Mon, 14 Apr 2025 08:11:25 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Uchida",
"Kengo",
""
],
[
"Shibuya",
"Takashi",
""
],
[
"Takida",
"Yuhta",
""
],
[
"Murata",
"Naoki",
""
],
[
"Tanke",
"Julian",
""
],
[
"Takahashi",
"Shusuke",
""
],
[
"Mitsufuji",
"Yuki",
""
]
] |
2406.02899 | Sylvain Fichet | Sergio Barbosa, Sylvain Fichet, Eugenio Megias, Mariano Quiros | Entanglement Entropy and Thermal Phase Transitions from Curvature
Singularities | 44 pages, 8 figures, 1 table. v2: Text revisions, appendix added,
references updated, matches JHEP version | J. High Energ. Phys. 2025, 44 (2025) | 10.1007/JHEP04(2025)044 | null | hep-th | http://creativecommons.org/licenses/by/4.0/ | We study holographic entanglement entropy and revisit thermodynamics and
confinement in the dilaton-gravity system. Our analysis focuses on a solvable
class of backgrounds that includes AdS and linear dilaton spacetimes as
particular cases, with some results extended to general warped metrics. A
general lesson is that the behavior of the holographic theory is tied to the
bulk curvature singularities. We find that a singular background is confining
if and only if i) the singularity coincides with a boundary or ii) it is the
linear dilaton. In the former case, for which the singularity cuts off
spacetime, we demonstrate that both entanglement entropy and thermodynamics
exhibit a first order phase transition. In the linear dilaton case we find
instead that both entanglement entropy and thermal phase transitions are of
second order. Additionally, along the process we thoroughly derive the radion
effective action at quadratic order.
| [
{
"version": "v1",
"created": "Wed, 5 Jun 2024 03:39:36 GMT"
},
{
"version": "v2",
"created": "Mon, 10 Mar 2025 11:46:05 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Barbosa",
"Sergio",
""
],
[
"Fichet",
"Sylvain",
""
],
[
"Megias",
"Eugenio",
""
],
[
"Quiros",
"Mariano",
""
]
] |
2406.04445 | Xu Chen | Xu Chen, Yuanjie Sun, Eugen Hruska, Vivek Dixit, Jinming Yang, Yu He,
Yao Wang, Fang Liu | Detecting Thermodynamic Phase Transition via Explainable Machine
Learning of Photoemission Spectroscopy | 10 pages, 6 figures | Newton 1, 100066 (2025) | 10.1016/j.newton.2025.100066 | null | cond-mat.supr-con cond-mat.str-el | http://creativecommons.org/licenses/by/4.0/ | Identifying thermodynamic signatures of electronic phases, such as
superconductivity, is challenging in low-dimensional materials due to strong
fluctuations and low probing volume. Spectroscopic methods are often used to
identify new bulk phases, but their main measurable quantity -- electronic
energy gaps -- is no longer an effective order parameter in low-dimensional and
fluctuating systems. Combining angle-resolved photoemission with a
domain-adversarial neural network, we report a data-driven method to identify
thermodynamic phase transitions solely based on single-particle spectra. We
demonstrate 97.6$\%$ accuracy in cuprate superconductor
Bi$_2$Sr$_2$CaCu$_2$O$_{8+\delta}$ with strong superconducting fluctuations.
This model notably compensates for the scarcity of experimental data by
leveraging virtually inexhaustible simulated data. Further, its explainability
reveals the crucial role of in-gap spectral weight in detecting phase
fluctuations and thermodynamic transitions. Our work pinpoints the
spectroscopic signatures of fluctuating orders and enables using spectroscopy
for machine-learning-assisted material discovery for low-dimensional and strong
coupling systems.
| [
{
"version": "v1",
"created": "Thu, 6 Jun 2024 19:00:30 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 15:12:47 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Chen",
"Xu",
""
],
[
"Sun",
"Yuanjie",
""
],
[
"Hruska",
"Eugen",
""
],
[
"Dixit",
"Vivek",
""
],
[
"Yang",
"Jinming",
""
],
[
"He",
"Yu",
""
],
[
"Wang",
"Yao",
""
],
[
"Liu",
"Fang",
""
]
] |
2406.04844 | Yuhao Li | Yuhao Li, Jiale Cao, Muzammal Naseer, Yu Zhu, Jinqiu Sun, Yanning
Zhang, Fahad Shahbaz Khan | Multi-Granularity Language-Guided Training for Multi-Object Tracking | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Most existing multi-object tracking methods typically learn visual tracking
features via maximizing dis-similarities of different instances and minimizing
similarities of the same instance. While such a feature learning scheme
achieves promising performance, learning discriminative features solely based
on visual information is challenging especially in case of environmental
interference such as occlusion, blur and domain variance. In this work, we
argue that multi-modal language-driven features provide complementary
information to classical visual features, thereby aiding in improving the
robustness to such environmental interference. To this end, we propose a new
multi-object tracking framework, named LG-MOT, that explicitly leverages
language information at different levels of granularity (scene-and
instance-level) and combines it with standard visual features to obtain
discriminative representations. To develop LG-MOT, we annotate existing MOT
datasets with scene-and instance-level language descriptions. We then encode
both instance-and scene-level language information into high-dimensional
embeddings, which are utilized to guide the visual features during training. At
inference, our LG-MOT uses the standard visual features without relying on
annotated language descriptions. Extensive experiments on three benchmarks,
MOT17, DanceTrack and SportsMOT, reveal the merits of the proposed
contributions leading to state-of-the-art performance. On the DanceTrack test
set, our LG-MOT achieves an absolute gain of 2.2\% in terms of target object
association (IDF1 score), compared to the baseline using only visual features.
Further, our LG-MOT exhibits strong cross-domain generalizability. The dataset
and code will be available at https://github.com/WesLee88524/LG-MOT.
| [
{
"version": "v1",
"created": "Fri, 7 Jun 2024 11:18:40 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 19:47:19 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Li",
"Yuhao",
""
],
[
"Cao",
"Jiale",
""
],
[
"Naseer",
"Muzammal",
""
],
[
"Zhu",
"Yu",
""
],
[
"Sun",
"Jinqiu",
""
],
[
"Zhang",
"Yanning",
""
],
[
"Khan",
"Fahad Shahbaz",
""
]
] |
2406.05067 | Kang Lu | Kang Lu, Weiqiang Wang, Weinan Zhang | Affine $\imath$quantum groups and twisted Yangians in Drinfeld
presentations | 33 pages; v2, mild modifications and updates, to appear in CMP | Comm. Math. Phys., 406 (2025), no. 5, article no. 98, 36 pp | 10.1007/s00220-025-05263-z | null | math.QA math-ph math.MP math.RT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We formulate a family of algebras, twisted Yangians (of split type) in
current generators and relations, via a degeneration of the Drinfeld
presentation of affine $\imath$quantum groups (associated with split Satake
diagrams). These new algebras admit PBW type bases and are shown to be a
deformation of twisted current algebras; presentations for twisted current
algebras are also provided. For type AI, it matches with the Drinfeld
presentation of twisted Yangian obtained via Gauss decomposition. We conjecture
that our split twisted Yangians are isomorphic to the corresponding ones in RTT
presentation.
| [
{
"version": "v1",
"created": "Fri, 7 Jun 2024 16:37:20 GMT"
},
{
"version": "v2",
"created": "Tue, 4 Feb 2025 20:13:31 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Lu",
"Kang",
""
],
[
"Wang",
"Weiqiang",
""
],
[
"Zhang",
"Weinan",
""
]
] |
2406.06044 | Sunjae Yoon | Sunjae Yoon, Gwanhyeong Koo, Geonwoo Kim, Chang D. Yoo | FRAG: Frequency Adapting Group for Diffusion Video Editing | 16 pages, 16 figures, ICML 2024 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In video editing, the hallmark of a quality edit lies in its consistent and
unobtrusive adjustment. Modification, when integrated, must be smooth and
subtle, preserving the natural flow and aligning seamlessly with the original
vision. Therefore, our primary focus is on overcoming the current challenges in
high quality edit to ensure that each edit enhances the final product without
disrupting its intended essence. However, quality deterioration such as
blurring and flickering is routinely observed in recent diffusion video editing
systems. We confirm that this deterioration often stems from high-frequency
leak: the diffusion model fails to accurately synthesize high-frequency
components during denoising process. To this end, we devise Frequency Adapting
Group (FRAG) which enhances the video quality in terms of consistency and
fidelity by introducing a novel receptive field branch to preserve
high-frequency components during the denoising process. FRAG is performed in a
model-agnostic manner without additional training and validates the
effectiveness on video editing benchmarks (i.e., TGVE, DAVIS).
| [
{
"version": "v1",
"created": "Mon, 10 Jun 2024 06:24:19 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 11:23:41 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Yoon",
"Sunjae",
""
],
[
"Koo",
"Gwanhyeong",
""
],
[
"Kim",
"Geonwoo",
""
],
[
"Yoo",
"Chang D.",
""
]
] |
2406.07062 | Pierre-Fran\c{c}ois Loos Dr | Antoine Marie and Pina Romaniello and Pierre-Fran\c{c}ois Loos | Anomalous propagators and the particle-particle channel: Hedin's
equations | 12 pages, 5 figures (Supp. Mat. available) | Phys. Rev. B 110, 115155 (2024) | 10.1103/PhysRevB.110.115155 | null | physics.chem-ph cond-mat.mtrl-sci cond-mat.str-el nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Hedin's equations provide an elegant route to compute the exact one-body
Green's function (or propagator) via the self-consistent iteration of a set of
non-linear equations.
Its first-order approximation, known as $GW$, corresponds to a resummation of
ring diagrams and has shown to be extremely successful in physics and
chemistry. Systematic improvement is possible, although challenging, via the
introduction of vertex corrections.
Considering anomalous propagators and an external pairing potential, we
derive a new self-consistent set of closed equations equivalent to the famous
Hedin equations but having as a first-order approximation the particle-particle
(pp) $T$-matrix approximation where one performs a resummation of the ladder
diagrams.
This pp version of Hedin's equations offers a way to go systematically beyond
the $T$-matrix approximation by accounting for low-order pp vertex corrections.
| [
{
"version": "v1",
"created": "Tue, 11 Jun 2024 08:44:52 GMT"
},
{
"version": "v2",
"created": "Wed, 4 Sep 2024 12:53:30 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Marie",
"Antoine",
""
],
[
"Romaniello",
"Pina",
""
],
[
"Loos",
"Pierre-François",
""
]
] |
2406.08169 | Salahuddin Abdul Rahman | Salahuddin Abdul Rahman, \"Ozkan Karabacak, Rafal Wisniewski | Feedback-Based Quantum Algorithm for Constrained Optimization Problems | This paper is accepted for publication in the 15th International
Conference on Parallel Processing and Applied Mathematics (PPAM 2024) | null | 10.1007/978-3-031-85700-3_20 | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | The feedback-based algorithm for quantum optimization (FALQON) has recently
been proposed to solve quadratic unconstrained binary optimization problems.
This paper efficiently generalizes FALQON to tackle quadratic constrained
binary optimization (QCBO) problems. For this purpose, we introduce a new
operator that encodes the problem's solution as its ground state. Using
Lyapunov control theory, we design a quantum control system such that the state
converges to the ground state of this operator. When applied to the QCBO
problem, we show that our proposed algorithm saves computational resources by
reducing the depth of the quantum circuit and can perform better than FALQON.
The effectiveness of our proposed algorithm is further illustrated through
numerical simulations.
| [
{
"version": "v1",
"created": "Wed, 12 Jun 2024 12:58:43 GMT"
},
{
"version": "v2",
"created": "Fri, 19 Jul 2024 09:21:41 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Rahman",
"Salahuddin Abdul",
""
],
[
"Karabacak",
"Özkan",
""
],
[
"Wisniewski",
"Rafal",
""
]
] |
2406.08315 | Luca Marzari | Luca Marzari, Priya L. Donti, Changliu Liu and Enrico Marchesini | Improving Policy Optimization via $\varepsilon$-Retrain | Accepted at AAMAS 2025 | null | null | null | cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present $\varepsilon$-retrain, an exploration strategy encouraging a
behavioral preference while optimizing policies with monotonic improvement
guarantees. To this end, we introduce an iterative procedure for collecting
retrain areas -- parts of the state space where an agent did not satisfy the
behavioral preference. Our method switches between the typical uniform restart
state distribution and the retrain areas using a decaying factor $\varepsilon$,
allowing agents to retrain on situations where they violated the preference. We
also employ formal verification of neural networks to provably quantify the
degree to which agents adhere to these behavioral preferences. Experiments over
hundreds of seeds across locomotion, power network, and navigation tasks show
that our method yields agents that exhibit significant performance and sample
efficiency improvements.
| [
{
"version": "v1",
"created": "Wed, 12 Jun 2024 15:16:26 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 14:36:00 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Marzari",
"Luca",
""
],
[
"Donti",
"Priya L.",
""
],
[
"Liu",
"Changliu",
""
],
[
"Marchesini",
"Enrico",
""
]
] |
2406.08826 | Natsuko Ishida | Natsuko Ishida, Motohiko Ezawa, Guangtai Lu, Wenbo Lin, Yasutomo Ota,
Yasuhiko Arakawa, and Satoshi Iwamoto | Topological Corner States in Bilayer and Trilayer Systems with
Vertically Stacked Topological Heterostructures | 13 pages, 9 figures | Phys. Rev. B 111, 115418 (2025) | 10.1103/PhysRevB.111.115418 | null | cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | We investigate bilayer and trilayer systems composed of topologically
distinct, vertically stacked layers, forming topological heterostructures based
on the Benalcazar-Bernevig-Hughes model. We find that a topological phase
transition induced by interlayer coupling significantly alters the number of
corner states in these topological structures. Furthermore, we find that
traditional nested Wilson loop analysis inaccurately classifies certain phases,
leading us to evaluate multipole chiral numbers (MCNs) as a more appropriate
topological invariant for this scenario. The MCNs not only enable accurate
classification of topological phases but also directly correspond to the number
of zero-energy corner states, effectively characterizing $\mathbb{Z}$-class
HOTI phases. Our study proposes the novel concept of topological
heterostructures, providing critical insights into the control of localized
corner states within multilayer systems and expanding potential research
directions.
| [
{
"version": "v1",
"created": "Thu, 13 Jun 2024 05:37:24 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 01:16:02 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ishida",
"Natsuko",
""
],
[
"Ezawa",
"Motohiko",
""
],
[
"Lu",
"Guangtai",
""
],
[
"Lin",
"Wenbo",
""
],
[
"Ota",
"Yasutomo",
""
],
[
"Arakawa",
"Yasuhiko",
""
],
[
"Iwamoto",
"Satoshi",
""
]
] |
2406.08947 | Peter Mlkvik | Lea Haas, Peter Mlkvik, Nicola A. Spaldin and Claude Ederer | Incorporating static intersite correlation effects in vanadium dioxide
through DFT$+V$ | null | Phys. Rev. Research 6, 043177 (2024) | 10.1103/PhysRevResearch.6.043177 | null | cond-mat.mtrl-sci cond-mat.str-el | http://creativecommons.org/licenses/by/4.0/ | We analyze the effects on the structural and electronic properties of
vanadium dioxide (VO$_2$) of adding an empirical inter-atomic potential within
the density-functional theory$+V$ (DFT$+V$) framework. We use the DFT$+V$
machinery founded on the extended Hubbard model to apply an empirical
self-energy correction between nearest-neighbor vanadium atoms in both rutile
and monoclinic phases, and for a set of structures interpolating between these
two cases. We observe that imposing an explicit intersite interaction $V$ along
the vanadium-vanadium chains enhances the characteristic bonding-antibonding
splitting of the relevant bands in the monoclinic phase, thus favoring
electronic dimerization and the formation of a band gap. We then explore the
effect of $V$ on the structural properties and the relative energies of the two
phases, finding an insulating global energy minimum for the monoclinic phase,
consistent with experimental observations. With increasing $V$, this minimum
becomes deeper relative to the rutile structure, and the transition from the
metallic to the insulating state becomes sharper. We also analyze the effect of
applying the $+V$ correction either to all or only to selected
vanadium-vanadium pairs, and both in the monoclinic as well as in the metallic
rutile phase. Our results suggest that DFT$+V$ can indeed serve as a
computationally inexpensive unbiased way of modeling VO$_2$ which is well
suited for studies that, e.g., require large system sizes.
| [
{
"version": "v1",
"created": "Thu, 13 Jun 2024 09:17:31 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Haas",
"Lea",
""
],
[
"Mlkvik",
"Peter",
""
],
[
"Spaldin",
"Nicola A.",
""
],
[
"Ederer",
"Claude",
""
]
] |
2406.09090 | Petru Jebelean | Petru Jebelean | Potential systems with singular $\Phi$-Laplacian | null | null | 10.1142/S0219199725500415 | null | math.AP math.CA | http://creativecommons.org/licenses/by/4.0/ | We are concerned with solvability of the boundary value problem $$-\left[
\phi(u^{\prime}) \right] ^{\prime}=\nabla_u F(t,u), \quad \left ( \phi \left(
u^{\prime }\right)(0), -\phi \left( u^{\prime }\right)(T)\right )\in \partial
j(u(0), u(T)),$$ where $\phi$ is a homeomorphism from $B_a$ -- the open ball of
radius $a$ centered at $0_{\mathbb{R}^N},$ onto $\mathbb{R}^N$, satisfying
$\phi(0_{\mathbb{R}^N})=0_{\mathbb{R}^N}$, $\phi =\nabla \Phi$, with $\Phi:
\overline{B}_a \to (-\infty, 0]$ of class $C^1$ on $B_a$, continuous and
strictly convex on $\overline{B}_a.$ The potential $F:[0,T] \times \mathbb{R}^N
\to \mathbb{R}$ is of class $C^1$ with respect to the second variable and
$j:\mathbb{R}^N \times \mathbb{R}^N \rightarrow (-\infty, +\infty]$ is proper,
convex and lower semicontinuous. We first provide a variational formulation in
the frame of critical point theory for convex, lower semicontinuous
perturbations of $C^1$-functionals. Then, taking the advantage of this key
step, we obtain existence of minimum energy as well as saddle-point solutions
of the problem. Some concrete illustrative examples of applications are
provided.
| [
{
"version": "v1",
"created": "Thu, 13 Jun 2024 13:18:21 GMT"
},
{
"version": "v2",
"created": "Thu, 9 Jan 2025 11:51:15 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Jebelean",
"Petru",
""
]
] |
2406.09233 | Luca Collini | Luca Collini, Siddharth Garg, Ramesh Karri | C2HLSC: Can LLMs Bridge the Software-to-Hardware Design Gap? | Accepted at The First IEEE International Workshop on LLM-Aided Design | null | 10.1109/LAD62341.2024.10691856 | null | cs.AR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | High Level Synthesis (HLS) tools offer rapid hardware design from C code, but
their compatibility is limited by code constructs. This paper investigates
Large Language Models (LLMs) for refactoring C code into HLS-compatible
formats. We present several case studies by using an LLM to rewrite C code for
NIST 800-22 randomness tests, a QuickSort algorithm and AES-128 into
HLS-synthesizable c. The LLM iteratively transforms the C code guided by user
prompts, implementing functions like streaming data and hardware-specific
signals. This evaluation demonstrates the LLM's potential to assist hardware
design refactoring regular C code into HLS synthesizable C code.
| [
{
"version": "v1",
"created": "Thu, 13 Jun 2024 15:33:54 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Collini",
"Luca",
""
],
[
"Garg",
"Siddharth",
""
],
[
"Karri",
"Ramesh",
""
]
] |
2406.09593 | John Cobb | John Cobb, Nathaniel Gallup, John Spoerl | Semigroup Graded Stillman's Conjecture | 7 pages | Journal of Algebra, 673 (2025) no. 1 | 10.1016/j.jalgebra.2025.02.016 | null | math.AC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We resolve Stillman's conjecture for families of polynomial rings that are
graded by any semigroup under mild conditions. Conversely, we show that these
conditions are necessary for the existence of a Stillman bound. This has
applications even for the well-known standard graded case.
| [
{
"version": "v1",
"created": "Thu, 13 Jun 2024 21:26:02 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 19:46:31 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Cobb",
"John",
""
],
[
"Gallup",
"Nathaniel",
""
],
[
"Spoerl",
"John",
""
]
] |
2406.10029 | Sarah Timhadjelt | Sarah Timhadjelt | Non-Hermitian expander obtained with Haar distributed unitaries | null | null | null | null | math.PR math-ph math.MP quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider a random quantum channel obtained by taking a selection of $d$
independent and Haar distributed $N$ dimensional unitaries. We follow the
argument of Hastings to bound the spectral gap in terms of eigenvalues and
adapt it to give an exact estimate of the spectral gap in terms of singular
values \cite{hastings2007random,harrow2007quantum}. This shows that we have
constructed a random quantum expander in terms of both singular values and
eigenvalues. The lower bound is an analog of the Alon-Boppana bound for
$d$-regular graphs. The upper bound is obtained using Schwinger-Dyson
equations.
| [
{
"version": "v1",
"created": "Fri, 14 Jun 2024 13:37:46 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 14:35:30 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Timhadjelt",
"Sarah",
""
]
] |
2406.10066 | Sanaz Mahmoodi Takaghaj | Sanaz Mahmoodi Takaghaj, Jack Sampson | D-SELD: Dataset-Scalable Exemplar LCA-Decoder | Accepted to Neuromorphic Computing and Engineering journal(NCE), 2024 | null | 10.1088/2634-4386/ad9e2c | null | cs.ET | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Neuromorphic computing has recently gained significant attention as a
promising approach for developing energy-efficient, massively parallel
computing systems inspired by the spiking behavior of the human brain and
natively mapping Spiking Neural Networks (SNNs). Effective training algorithms
for SNNs are imperative for increased adoption of neuromorphic platforms;
however, SNN training continues to lag behind advances in other classes of ANN.
In this paper, we reduce this gap by proposing an innovative encoder-decoder
technique that leverages sparse coding and the Locally Competitive Algorithm
(LCA) to provide an algorithm specifically designed for neuromorphic platforms.
Using our proposed Dataset-Scalable Exemplar LCA-Decoder we reduce the
computational demands and memory requirements associated with training SNNs
using error backpropagation methods on increasingly larger training sets. We
offer a solution that can be scalably applied to datasets of any size. Our
results show the highest reported top-1 test accuracy using SNNs on the
ImageNet and CIFAR100 datasets, surpassing previous benchmarks. Specifically,
we achieved a record top-1 accuracy of 80.75% on ImageNet (ILSVRC2012
validation set) and 79.32% on CIFAR100 using SNNs.
| [
{
"version": "v1",
"created": "Fri, 14 Jun 2024 14:23:35 GMT"
},
{
"version": "v2",
"created": "Thu, 31 Oct 2024 19:39:07 GMT"
},
{
"version": "v3",
"created": "Sun, 13 Apr 2025 01:38:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Takaghaj",
"Sanaz Mahmoodi",
""
],
[
"Sampson",
"Jack",
""
]
] |
2406.10244 | Sheng Zhang | Sheng Zhang and Maolin Wang and Wanyu Wang and Jingtong Gao and
Xiangyu Zhao and Yu Yang and Xuetao Wei and Zitao Liu and Tong Xu | GLINT-RU: Gated Lightweight Intelligent Recurrent Units for Sequential
Recommender Systems | null | null | null | null | cs.IR cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Transformer-based models have gained significant traction in sequential
recommender systems (SRSs) for their ability to capture user-item interactions
effectively. However, these models often suffer from high computational costs
and slow inference. Meanwhile, existing efficient SRS approaches struggle to
embed high-quality semantic and positional information into latent
representations. To tackle these challenges, this paper introduces GLINT-RU, a
lightweight and efficient SRS leveraging a single-layer dense selective Gated
Recurrent Units (GRU) module to accelerate inference. By incorporating a dense
selective gate, GLINT-RU adaptively captures temporal dependencies and
fine-grained positional information, generating high-quality latent
representations. Additionally, a parallel mixing block infuses fine-grained
positional features into user-item interactions, enhancing both recommendation
quality and efficiency. Extensive experiments on three datasets demonstrate
that GLINT-RU achieves superior prediction accuracy and inference speed,
outperforming baselines based on RNNs, Transformers, MLPs, and SSMs. These
results establish GLINT-RU as a powerful and efficient solution for SRSs.
| [
{
"version": "v1",
"created": "Thu, 6 Jun 2024 13:55:55 GMT"
},
{
"version": "v2",
"created": "Sat, 1 Feb 2025 15:23:41 GMT"
},
{
"version": "v3",
"created": "Sat, 12 Apr 2025 17:18:17 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhang",
"Sheng",
""
],
[
"Wang",
"Maolin",
""
],
[
"Wang",
"Wanyu",
""
],
[
"Gao",
"Jingtong",
""
],
[
"Zhao",
"Xiangyu",
""
],
[
"Yang",
"Yu",
""
],
[
"Wei",
"Xuetao",
""
],
[
"Liu",
"Zitao",
""
],
[
"Xu",
"Tong",
""
]
] |
2406.10329 | Fabio Pacucci | Emmanuel Durodola, Fabio Pacucci, Ryan C. Hickox | Exploring the AGN Fraction of a Sample of JWST's Little Red Dots at $5 <
z < 8$: Overmassive Black Holes Are Strongly Favored | Accepted for publication in The Astrophysical Journal. This is the
final version. 14 pages, 6 figures | null | null | null | astro-ph.GA astro-ph.CO astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | JWST is revolutionizing our view of the early Universe by pushing the
boundaries of detectable galaxies and black holes in redshift (upward) and mass
(downward). The Little Red Dots (LRDs), detected by several surveys at $z > 4$,
present a significant interpretational challenge, as their Spectral Energy
Distributions (SED) can mimic both AGN and stellar population templates. This
study analyzes 19 LRDs from the JADES survey, utilizing NIRCam and MIRI
photometry. By performing SED fitting across a vast parameter space, we explore
a broad range of AGN fractions, defined as the ratio of the monochromatic
luminosities (AGN, galaxy, and dust) over a specified wavelength range, 0.4 -
0.7 $\mu m$ rest-frame. We find that 17 of the 19 LRDs investigated are
consistent with having significant AGN contributions, with best-fitting AGN
fractions ranging between 20% and 70%, while one galaxy shows a low AGN
contribution (2%) and another appears to be purely star-forming. Moreover,
assuming these LRDs do indeed host AGN, we can place limits on their black hole
masses using the inferred AGN bolometric luminosities and adopting the
Eddington limit. We find that, independent of the specific AGN fraction
adopted, the LRDs' black holes are significantly overmassive relative to their
host galaxies (by $\sim 1$ dex, and up to $\sim 4$ dex in the most extreme
cases) compared to the local $M_{\bullet} - M_{\star}$ relation. The presence
of overmassive black holes in the high-$z$ Universe may provide the strongest
evidence yet of heavy black hole seeding occurring during the cosmic dark ages.
| [
{
"version": "v1",
"created": "Fri, 14 Jun 2024 18:00:00 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Jan 2025 17:10:39 GMT"
},
{
"version": "v3",
"created": "Mon, 14 Apr 2025 13:58:08 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Durodola",
"Emmanuel",
""
],
[
"Pacucci",
"Fabio",
""
],
[
"Hickox",
"Ryan C.",
""
]
] |
2406.10344 | Sasanka Dowarah | Sasanka Dowarah, Chuanwei Zhang, Vedika Khemani, Michael H.
Kolodrubetz | Phases and phase transition in Grover's algorithm with systematic noise | 14 pages, 11 figures | Phys. Rev. A 111.042603 (2025) | 10.1103/PhysRevA.111.042603 | null | quant-ph cond-mat.stat-mech | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While limitations on quantum computation by Markovian environmental noise are
well-understood in generality, their behavior for different quantum circuits
and noise realizations can be less universal. Here we consider a canonical
quantum algorithm - Grover's algorithm for unordered search on $L$ qubits - in
the presence of systematic noise. This allows us to write the behavior as a
random Floquet unitary, which we show is well-characterized by random matrix
theory (RMT). The RMT analysis enables analytical predictions for phases and
phase transitions of the many-body dynamics. We find two separate transitions.
At moderate disorder $\delta_{c,\mathrm{gap}}\sim L^{-1}$, there is a
ergodicity breaking transition such that a finite-dimensional manifold remains
non-ergodic for $\delta < \delta_{c,\mathrm{gap}}$. Computational power is lost
at a much smaller disorder, $\delta_{c,\mathrm{comp}} \sim L^{-1/2}2^{-L/2}$.
We comment on relevance to non-systematic noise in realistic quantum computers,
including cold atom, trapped ion, and superconducting platforms.
| [
{
"version": "v1",
"created": "Fri, 14 Jun 2024 18:00:06 GMT"
},
{
"version": "v2",
"created": "Thu, 3 Apr 2025 17:15:27 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Dowarah",
"Sasanka",
""
],
[
"Zhang",
"Chuanwei",
""
],
[
"Khemani",
"Vedika",
""
],
[
"Kolodrubetz",
"Michael H.",
""
]
] |
2406.10369 | Andreas Pape | Andreas Duus Pape, J. David Schaffer, Hiroki Sayama, Christopher Zosh | On the Preservation of Input/Output Directed Graph Informativeness under
Crossover | null | null | null | null | cs.SI | http://creativecommons.org/licenses/by/4.0/ | There is a broad class of networks which connect inputs to outputs. We
provide a strong theoretical foundation for crossover across this class and
connect it to informativeness, a measure of the connectedness of inputs to
outputs. We define Input/Output Directed Graphs (or IOD Graphs) as graphs with
nodes $N$ and directed edges $E$, where $N$ contains (a) a set of "input nodes"
$I \subset N$, where each $i \in I$ has no incoming edges and any number of
outgoing edges, and (b) a set of "output nodes" $O \subset N$, where each $o
\in O$ has no outgoing edges and any number of incoming edges, and $I\cap O =
\emptyset$. We define informativeness, which involves the connections via
directed paths from the input nodes to the output nodes: A partially
informative IOD Graph has at least one path from an input to an output, a very
informative IOD Graph has a path from every input to some output, and a fully
informative IOD Graph has a path from every input to every output.
A perceptron is an example of an IOD Graph. If it has non-zero weights and
any number of layers, it is fully informative. As links are removed (assigned
zero weight), the perceptron might become very, partially, or not informative.
We define a crossover operation on IOD Graphs in which we find subgraphs with
matching sets of forward and backward directed links to "swap." With this
operation, IOD Graphs can be subject to evolutionary computation methods. We
show that fully informative parents may yield a non-informative child. We also
show that under conditions of contiguousness and the no dangling nodes
condition, crossover compatible, partially informative parents yield partially
informative children, and very informative input parents with partially
informative output parents yield very informative children. However, even under
these conditions, full informativeness may not be retained.
| [
{
"version": "v1",
"created": "Fri, 14 Jun 2024 18:56:00 GMT"
},
{
"version": "v2",
"created": "Mon, 29 Jul 2024 16:47:40 GMT"
},
{
"version": "v3",
"created": "Mon, 14 Apr 2025 17:46:31 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Pape",
"Andreas Duus",
""
],
[
"Schaffer",
"J. David",
""
],
[
"Sayama",
"Hiroki",
""
],
[
"Zosh",
"Christopher",
""
]
] |
2406.10999 | Liman Wang | Hanyang Zhong, Liman Wang, Wenting Cao, Zeyuan Sun | Balancing Rigor and Utility: Mitigating Cognitive Biases in Large
Language Models for Multiple-Choice Questions | This work has been accepted as a full paper at the 2025 Annual
Conference of the Cognitive Science Society (CogSci 2025) and will be
presented in the form of a poster. The dataset and project website are
available at: https://hanyangzhong.github.io/BRU-website/ | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper examines the role of cognitive biases in the decision-making
processes of large language models (LLMs), challenging the conventional goal of
eliminating all biases. When properly balanced, we show that certain cognitive
biases can enhance decision-making efficiency through rational deviations and
heuristic shortcuts. By introducing heuristic moderation and an abstention
option, which allows LLMs to withhold responses when uncertain, we reduce error
rates, improve decision accuracy, and optimize decision rates. Using the
Balance Rigor and Utility (BRU) dataset, developed through expert
collaboration, our findings demonstrate that targeted inspection of cognitive
biases aligns LLM decisions more closely with human reasoning, enhancing
reliability and suggesting strategies for future improvements. This approach
offers a novel way to leverage cognitive biases to improve the practical
utility of LLMs across various applications.
| [
{
"version": "v1",
"created": "Sun, 16 Jun 2024 16:25:22 GMT"
},
{
"version": "v2",
"created": "Mon, 2 Sep 2024 20:26:30 GMT"
},
{
"version": "v3",
"created": "Mon, 9 Sep 2024 16:28:09 GMT"
},
{
"version": "v4",
"created": "Tue, 8 Apr 2025 23:59:08 GMT"
},
{
"version": "v5",
"created": "Sun, 13 Apr 2025 13:03:09 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhong",
"Hanyang",
""
],
[
"Wang",
"Liman",
""
],
[
"Cao",
"Wenting",
""
],
[
"Sun",
"Zeyuan",
""
]
] |
2406.12681 | Anders W. Sandvik | Anders W. Sandvik | Using operator covariance to disentangle scaling dimensions in lattice
models | 21 pages, 16 figures. v2: Considerably expanded | null | null | null | cond-mat.stat-mech cond-mat.str-el hep-lat | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In critical lattice models, distance ($r$) dependent correlation functions
contain power laws $r^{-2\Delta}$ governed by scaling dimensions $\Delta$ of an
underlying continuum field theory. In Monte Carlo simulations, the leading
dimensions can be extracted by data fitting, which is difficult when two or
more powers contribute significantly. Here a method utilizing covariance
between multiple lattice operators is developed where the $r$ dependent
eigenvalues of the covariance matrix reflect scaling dimensions of individual
field operators. This disentangling is demonstrated explicitly for conformal
field theories. The scheme is first tested on the critical point of the 2D
Ising model, where the two primary scaling dimensions and their respective two
lowest descendant dimensions are extracted. The 3D Ising model is studied next,
revealing the two relevant primaries and their lowest descendants to high
precision. The 2D tricritical Ising point is studied with the Blume-Capel
model. Here the scaling dimensions of all three symmetric primary operators are
successfully isolated along with the leading descendants. The eigenvectors are
also studied and give useful information on the boundary between the ordered
and disordered phases in the neighborhood of the tricritical point. Finally,
the crossover from regular to tricritical Ising scaling is investigated on
several points on the phase boundary of the Blume-Capel model away from its
tricritical point. The scaling of the eigenvalues corresponding to tricritical
descendant operators are found to be remarkably stable even far from the
tricritical point. The covariance method represents a simple extension of
standard analysis of correlation functions and can significantly enhance the
utility of Monte Carlo simulations and other computational methods in studies
of criticality, in particular conformal critical points.
| [
{
"version": "v1",
"created": "Tue, 18 Jun 2024 14:52:07 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 07:42:06 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Sandvik",
"Anders W.",
""
]
] |
2406.13073 | Md Hasan Shahriar | Md Hasan Shahriar, Ning Wang, Naren Ramakrishnan, Y. Thomas Hou and
Wenjing Lou | Let the Noise Speak: Harnessing Noise for a Unified Defense Against
Adversarial and Backdoor Attacks | 20 pages, 9 figures | null | null | null | cs.LG cs.CR cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The exponential adoption of machine learning (ML) is propelling the world
into a future of distributed and intelligent automation and data-driven
solutions. However, the proliferation of malicious data manipulation attacks
against ML, namely adversarial and backdoor attacks, jeopardizes its
reliability in safety-critical applications. The existing detection methods are
attack-specific and built upon some strong assumptions, limiting them in
diverse practical scenarios. Thus, motivated by the need for a more robust,
unified, and attack-agnostic defense mechanism, we first investigate the shared
traits of adversarial and backdoor attacks. Based on our observation, we
propose NoiSec, a reconstruction-based intrusion detection system that brings a
novel perspective by shifting focus from the reconstructed input to the
reconstruction noise itself, which is the foundational root cause of such
malicious data alterations. NoiSec disentangles the noise from the test input,
extracts the underlying features from the noise, and leverages them to
recognize systematic malicious manipulation. Our comprehensive evaluation of
NoiSec demonstrates its high effectiveness across various datasets, including
basic objects, natural scenes, traffic signs, medical images, spectrogram-based
audio data, and wireless sensing against five state-of-the-art adversarial
attacks and three backdoor attacks under challenging evaluation conditions.
NoiSec demonstrates strong detection performance in both white-box and
black-box adversarial attack scenarios, significantly outperforming the closest
baseline models, particularly in an adaptive attack setting. We will provide
the code for future baseline comparison. Our code and artifacts are publicly
available at https://github.com/shahriar0651/NoiSec.
| [
{
"version": "v1",
"created": "Tue, 18 Jun 2024 21:44:51 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 03:19:46 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Shahriar",
"Md Hasan",
""
],
[
"Wang",
"Ning",
""
],
[
"Ramakrishnan",
"Naren",
""
],
[
"Hou",
"Y. Thomas",
""
],
[
"Lou",
"Wenjing",
""
]
] |
2406.13192 | Nadiia Derevianko | Nadiia Derevianko | Recovery of rational functions via Hankel pencil method and
sensitivities of the poles | 23 pages | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the paper, we develop a new method for the recovery of rational functions.
Our idea is based on the property that Fourier coefficients of rational
functions have the exponential structure and reconstruction of this exponential
structure with the ESPRIT method in the frequency domain. Further we present
sensitivity analysis for poles of rational functions reconstructed with our
method in case of unstructured and structured perturbations. Finally, we
consider several numerical experiments and, using sensitivities, explain the
recovery errors for poles.
| [
{
"version": "v1",
"created": "Wed, 19 Jun 2024 03:49:31 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 14:24:18 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Derevianko",
"Nadiia",
""
]
] |
2406.13690 | Pabitra Tripathy | Pabitra Tripathy | Lower bound of black hole hair in pure Lovelock theory of gravity | 6 pages, 1 figure. This version matches the published version | International Journal of Modern Physics D, 34, (2025), 2550017 | 10.1142/S0218271825500178 | null | gr-qc | http://creativecommons.org/licenses/by/4.0/ | As an alternative to the "no hair conjecture," the "no short hair conjecture"
for hairy black holes was established earlier. This theorem stipulates that
hair must be present above 3/2 of the event horizon radius for a hairy black
hole. It is assumed that the nonlinear behavior of the matter field plays a key
role in the presence of such hair. Subsequently, it was established that the
hair must extend beyond the photon sphere of the corresponding black hole. We
have investigated the validity of the "no short hair conjecture" in pure
Lovelock gravity. Our analysis has shown that irrespective of dimensionality
and Lovelock order, the hair of a static, spherically symmetric black hole
extends at least up to the photon sphere.
| [
{
"version": "v1",
"created": "Wed, 19 Jun 2024 16:42:13 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 12:06:38 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Tripathy",
"Pabitra",
""
]
] |
2406.14510 | Rotem Shalev-Arkushin | Rotem Shalev-Arkushin, Aharon Azulay, Tavi Halperin, Eitan Richardson,
Amit H. Bermano, Ohad Fried | V-LASIK: Consistent Glasses-Removal from Videos Using Synthetic Data | null | null | null | null | cs.CV cs.AI cs.GR | http://creativecommons.org/licenses/by/4.0/ | Diffusion-based generative models have recently shown remarkable image and
video editing capabilities. However, local video editing, particularly removal
of small attributes like glasses, remains a challenge. Existing methods either
alter the videos excessively, generate unrealistic artifacts, or fail to
perform the requested edit consistently throughout the video. In this work, we
focus on consistent and identity-preserving removal of glasses in videos, using
it as a case study for consistent local attribute removal in videos. Due to the
lack of paired data, we adopt a weakly supervised approach and generate
synthetic imperfect data, using an adjusted pretrained diffusion model. We show
that despite data imperfection, by learning from our generated data and
leveraging the prior of pretrained diffusion models, our model is able to
perform the desired edit consistently while preserving the original video
content. Furthermore, we exemplify the generalization ability of our method to
other local video editing tasks by applying it successfully to facial
sticker-removal. Our approach demonstrates significant improvement over
existing methods, showcasing the potential of leveraging synthetic data and
strong video priors for local video editing tasks.
| [
{
"version": "v1",
"created": "Thu, 20 Jun 2024 17:14:43 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 08:10:45 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Shalev-Arkushin",
"Rotem",
""
],
[
"Azulay",
"Aharon",
""
],
[
"Halperin",
"Tavi",
""
],
[
"Richardson",
"Eitan",
""
],
[
"Bermano",
"Amit H.",
""
],
[
"Fried",
"Ohad",
""
]
] |
2406.14533 | Christoph Minz | Christoph Minz | Local symmetries in partially ordered sets | 33 pages, 6 figures, 3 tables | null | null | null | math.CO gr-qc math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Partially ordered sets (posets) play a universal role as an abstract
structure in many areas of mathematics. For finite posets, an explicit
enumeration of distinct partial orders on a set of unlabelled elements is known
only up to a cardinality of 16 (listed as sequence A000112 in the OEIS), but
closed expressions are unknown. By considering the automorphisms of (finite)
posets, I introduce a formulation of local symmetries. These symmetries give
rise to a division operation on the set of posets and lead to the construction
of symmetry classes that are easier to characterise and enumerate. Furthermore,
we consider polynomial expressions that count certain subsets of posets with a
large number of layers (a large height). As an application in physics, local
symmetries or rather their absence helps to distinguish causal sets (locally
finite posets) that serve as discrete spacetime models from generic causal
sets.
| [
{
"version": "v1",
"created": "Thu, 20 Jun 2024 17:47:16 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 15:59:44 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Minz",
"Christoph",
""
]
] |
2406.15557 | Yuxuan Zhang | Yuxuan Zhang, Juan Carrasquilla, and Yong Baek Kim | Observation of a non-Hermitian supersonic mode on a trapped-ion quantum
computer | null | Nat Commun 16, 3286 (2025) | 10.1038/s41467-025-57930-3 | null | quant-ph cond-mat.str-el | http://creativecommons.org/licenses/by/4.0/ | Quantum computers have long been anticipated to excel in simulating quantum
many-body physics. While most previous work has focused on Hermitian physics,
we demonstrate the power of variational quantum circuits for resource-efficient
simulations of dynamical and equilibrium physics in non-Hermitian systems,
revealing new phenomena beyond standard Hermitian quantum machines. Using a
variational quantum compilation scheme for fermionic systems, we reduce gate
count, save qubits, and eliminate the need for postselection, a major challenge
in simulating non-Hermitian dynamics via standard Trotterization.
Experimentally, we observed a supersonic mode in the connected density-density
correlation function on an $ n = 18 $ fermionic chain after a non-Hermitian,
locally interacting quench, which would otherwise be forbidden by the
Lieb-Robinson bound in a Hermitian system. Additionally, we investigate
sequential quantum circuits generated by tensor networks for ground state
preparation, here defined as the eigenstate with the lowest real part
eigenvalue, using a variance minimization scheme. Through a trapped-ion
implementation on the Quantinuum H1 quantum processor, we accurately capture
correlation functions and energies across an exceptional point on a dissipative
spin chain up to length $ n = 20 $ using only 3 qubits. Motivated by these
advancements, we provide an analytical example demonstrating that simulating
single-qubit non-Hermitian dynamics for $\Theta(\log(n))$ time from certain
initial states is exponentially hard on a quantum computer, offering insights
into the opportunities and limitations of using quantum computation for
simulating non-Hermitian physics.
| [
{
"version": "v1",
"created": "Fri, 21 Jun 2024 18:00:06 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 18:52:13 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhang",
"Yuxuan",
""
],
[
"Carrasquilla",
"Juan",
""
],
[
"Kim",
"Yong Baek",
""
]
] |
2406.15583 | Hillary Dawkins | Kathleen C. Fraser, Hillary Dawkins, Svetlana Kiritchenko | Detecting AI-Generated Text: Factors Influencing Detectability with
Current Methods | null | Journal of Artificial Intelligence Research Vol. 82 (2025)
2233-2278 | 10.1613/jair.1.16665 | null | cs.CL cs.CY | http://creativecommons.org/licenses/by/4.0/ | Large language models (LLMs) have advanced to a point that even humans have
difficulty discerning whether a text was generated by another human, or by a
computer. However, knowing whether a text was produced by human or artificial
intelligence (AI) is important to determining its trustworthiness, and has
applications in many domains including detecting fraud and academic dishonesty,
as well as combating the spread of misinformation and political propaganda. The
task of AI-generated text (AIGT) detection is therefore both very challenging,
and highly critical. In this survey, we summarize state-of-the art approaches
to AIGT detection, including watermarking, statistical and stylistic analysis,
and machine learning classification. We also provide information about existing
datasets for this task. Synthesizing the research findings, we aim to provide
insight into the salient factors that combine to determine how "detectable"
AIGT text is under different scenarios, and to make practical recommendations
for future work towards this significant technical and societal challenge.
| [
{
"version": "v1",
"created": "Fri, 21 Jun 2024 18:31:49 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 17:42:35 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Fraser",
"Kathleen C.",
""
],
[
"Dawkins",
"Hillary",
""
],
[
"Kiritchenko",
"Svetlana",
""
]
] |
2406.15961 | Angeline Aguinaldo | Angeline Aguinaldo, Evan Patterson, William Regli | Automating Transfer of Robot Task Plans using Functorial Data Migrations | null | null | null | null | cs.RO cs.AI math.CT | http://creativecommons.org/licenses/by/4.0/ | This paper introduces a novel approach to ontology-based robot plan transfer
by leveraging functorial data migrations, a structured mapping method derived
from category theory. Functors provide structured maps between planning domain
ontologies which enables the transfer of task plans without the need for
replanning. Unlike methods tailored to specific plans, our framework applies
universally within the source domain once a structured map is defined. We
demonstrate this approach by transferring a task plan from the canonical
Blocksworld domain to one compatible with the AI2-THOR Kitchen environment.
Additionally, we discuss practical limitations, propose benchmarks for
evaluating symbolic plan transfer methods, and outline future directions for
scaling this approach.
| [
{
"version": "v1",
"created": "Sat, 22 Jun 2024 23:35:32 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 21:06:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Aguinaldo",
"Angeline",
""
],
[
"Patterson",
"Evan",
""
],
[
"Regli",
"William",
""
]
] |
2406.16222 | Christopher Kuo | Laurent C\^ot\'e, Christopher Kuo, David Nadler, and Vivek Shende | The microlocal Riemann-Hilbert correspondence for complex contact
manifolds | 67 pages | null | null | null | math.SG math.AG math.RT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Kashiwara showed in 1996 that the categories of microlocalized D-modules can
be canonically glued to give a sheaf of categories over a complex contact
manifold. Much more recently, and by rather different considerations, we
constructed a canonical notion of perverse microsheaves on the same class of
spaces. Here we provide a Riemann-Hilbert correspondence.
| [
{
"version": "v1",
"created": "Sun, 23 Jun 2024 21:31:08 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 21:41:39 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Côté",
"Laurent",
""
],
[
"Kuo",
"Christopher",
""
],
[
"Nadler",
"David",
""
],
[
"Shende",
"Vivek",
""
]
] |
2406.16843 | Arne Hole | Arne Hole | Constructibility, computational complexity and P versus NP | Compared to the previous version, there are minor changes only. Some
misprints have been corrected, the wording of the closing section and the
abstract has been changed, and some adjustments have been done | null | null | null | cs.CC cs.LO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A decision problem called the IPL problem is defined, and it is argued for
the validity of an associated thesis called the IPL thesis. This thesis states
that for some instances of the IPL problem, while an algorithm for verifying
correct solutions to the problem in polynomial time is explicitly
constructible, the IPL problem itself is algorithmically unsolvable in the
sense that no explicitly constructed algorithm can be verified as solving the
problem. Thus under a constructive interpretation of algorithmic complexity
classes, which is arguably their only meaningful interpretation, the IPL thesis
implies that NP is not contained in any complexity class consisting of
algorithmically solvable problems. In particular, the thesis implies a solution
to the P versus NP problem: P is not equal to NP. It also implies that NP is
not contained in larger complexity classes such as EXPTIME, seemingly
contradicting known results. However, the proofs of the these results all
tacitly employ an existence assumption which according to the IPL thesis, does
not hold for the IPL problem under a constructive interpretation of algorithmic
complexity classes.
| [
{
"version": "v1",
"created": "Mon, 24 Jun 2024 17:48:29 GMT"
},
{
"version": "v2",
"created": "Mon, 4 Nov 2024 09:36:14 GMT"
},
{
"version": "v3",
"created": "Mon, 11 Nov 2024 10:56:03 GMT"
},
{
"version": "v4",
"created": "Fri, 4 Apr 2025 08:53:34 GMT"
},
{
"version": "v5",
"created": "Mon, 14 Apr 2025 06:45:33 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Hole",
"Arne",
""
]
] |
2406.17871 | Di-De Yen | Michael Benedikt and Anthony Widjaja Lin and Di-De Yen | Revisiting the Expressiveness Landscape of Data Graph Queries | null | null | null | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The study of graph queries in database theory has spanned more than three
decades, resulting in a multitude of proposals for graph query languages. These
languages differ in the mechanisms. We can identify three main families of
languages, with the canonical representatives being: (1) regular path queries,
(2) walk logic, and (3) first-order logic with transitive closure operators.
This paper provides a complete picture of the expressive power of these
languages in the context of data graphs. Specifically, we consider a graph data
model that supports querying over both data and topology. For example, "Does
there exist a path between two different persons in a social network with the
same last name?". We also show that an extension of (1), augmented with
transitive closure operators, can unify the expressivity of (1)--(3) without
increasing the query evaluation complexity.
| [
{
"version": "v1",
"created": "Tue, 25 Jun 2024 18:19:14 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 10:43:43 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Benedikt",
"Michael",
""
],
[
"Lin",
"Anthony Widjaja",
""
],
[
"Yen",
"Di-De",
""
]
] |
2406.18145 | Shaowei Wang | Shaowei Wang, Changyu Dong, Xiangfu Song, Jin Li, Zhili Zhou, Di Wang,
Han Wu | Beyond Statistical Estimation: Differentially Private Individual
Computation via Shuffling | USENIX Security 2025. Full version with corrected consts in Theorem
5.2 and Appendix A | null | null | null | cs.CR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In data-driven applications, preserving user privacy while enabling valuable
computations remains a critical challenge. Technologies like differential
privacy have been pivotal in addressing these concerns. The shuffle model of DP
requires no trusted curators and can achieve high utility by leveraging the
privacy amplification effect yielded from shuffling. These benefits have led to
significant interest in the shuffle model. However, the computation tasks in
the shuffle model are limited to statistical estimation, making it inapplicable
to real-world scenarios in which each user requires a personalized output. This
paper introduces a novel paradigm termed Private Individual Computation (PIC),
expanding the shuffle model to support a broader range of
permutation-equivariant computations. PIC enables personalized outputs while
preserving privacy, and enjoys privacy amplification through shuffling. We
propose a concrete protocol that realizes PIC. By using one-time public keys,
our protocol enables users to receive their outputs without compromising
anonymity, which is essential for privacy amplification. Additionally, we
present an optimal randomizer, the Minkowski Response, designed for the PIC
model to enhance utility. We formally prove the security and privacy properties
of the PIC protocol. Theoretical analysis and empirical evaluations demonstrate
PIC's capability in handling non-statistical computation tasks, and the
efficacy of PIC and the Minkowski randomizer in achieving superior utility
compared to existing solutions.
| [
{
"version": "v1",
"created": "Wed, 26 Jun 2024 07:53:48 GMT"
},
{
"version": "v2",
"created": "Fri, 12 Jul 2024 01:36:06 GMT"
},
{
"version": "v3",
"created": "Thu, 30 Jan 2025 02:54:49 GMT"
},
{
"version": "v4",
"created": "Mon, 14 Apr 2025 06:19:02 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Wang",
"Shaowei",
""
],
[
"Dong",
"Changyu",
""
],
[
"Song",
"Xiangfu",
""
],
[
"Li",
"Jin",
""
],
[
"Zhou",
"Zhili",
""
],
[
"Wang",
"Di",
""
],
[
"Wu",
"Han",
""
]
] |
2406.18739 | Piotr Gai\'nski | Piotr Gai\'nski, Micha{\l} Koziarski, Krzysztof Maziarz, Marwin
Segler, Jacek Tabor, Marek \'Smieja | RetroGFN: Diverse and Feasible Retrosynthesis using GFlowNets | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Single-step retrosynthesis aims to predict a set of reactions that lead to
the creation of a target molecule, which is a crucial task in molecular
discovery. Although a target molecule can often be synthesized with multiple
different reactions, it is not clear how to verify the feasibility of a
reaction, because the available datasets cover only a tiny fraction of the
possible solutions. Consequently, the existing models are not encouraged to
explore the space of possible reactions sufficiently. In this paper, we propose
a novel single-step retrosynthesis model, RetroGFN, that can explore outside
the limited dataset and return a diverse set of feasible reactions by
leveraging a feasibility proxy model during the training. We show that RetroGFN
achieves competitive results on standard top-k accuracy while outperforming
existing methods on round-trip accuracy. Moreover, we provide empirical
arguments in favor of using round-trip accuracy, which expands the notion of
feasibility with respect to the standard top-k accuracy metric.
| [
{
"version": "v1",
"created": "Wed, 26 Jun 2024 20:10:03 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 12:15:24 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Gaiński",
"Piotr",
""
],
[
"Koziarski",
"Michał",
""
],
[
"Maziarz",
"Krzysztof",
""
],
[
"Segler",
"Marwin",
""
],
[
"Tabor",
"Jacek",
""
],
[
"Śmieja",
"Marek",
""
]
] |
2407.01164 | Gianluca Paolini | Simon Andr\'e and Gianluca Paolini | Around first-order rigidity of Coxeter groups | null | null | null | null | math.GR math.LO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | By the work of Sela, for any free group $F$, the Coxeter group $W_ 3 =
\mathbb{Z}/2\mathbb{Z} \ast \mathbb{Z}/2\mathbb{Z} \ast \mathbb{Z}/2\mathbb{Z}$
is elementarily equivalent to $W_3 \ast F$, and so Coxeter groups are not
closed under elementary equivalence among finitely generated groups. In this
paper we show that if we restrict to models which are generated by finitely
many torsion elements (finitely torsion-generated), then we can recover
striking rigidity results. Our main result is that if $(W, S)$ is a Coxeter
system whose irreducible components are either spherical, or affine or (Gromov)
hyperbolic, and $G$ is finitely torsion-generated and elementarily equivalent
to $W$, then $G$ is itself a Coxeter group. This combines results of the second
author et al. from [MPS22, PS23] with the following main hyperbolic result: if
$W$ is a Coxeter hyperbolic group and $G$ is $\mathrm{AE}$-equivalent to $W$
and finitely torsion-generated, then $G$ belongs to a finite collection of
Coxeter groups (modulo isomorphism). Furthermore, we show that there are two
hyperbolic Coxeter groups $W$ and $W'$ which are non-isomorphic but
$\mathrm{AE}$-equivalent. We also show that, on other hand, if we restrict to
certain specific classes of Coxeter groups then we can recover the strongest
possible form of first-order rigidity, which we call first-order
torsion-rigidity, namely the Coxeter group $W$ is the only finitely
torsion-generated model of its theory. Crucially, we show that this form of
rigidity holds for the following classes of Coxeter groups: even hyperbolic
Coxeter groups and free products of one-ended or finite hyperbolic Coxeter
groups. We conjecture that the same kind of phenomena occur for the whole class
of Coxeter groups. In this direction, we prove that if $W$ and $W'$ are even
Coxeter groups which are elementarily equivalent, then they are isomorphic.
| [
{
"version": "v1",
"created": "Mon, 1 Jul 2024 10:34:53 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 08:36:12 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"André",
"Simon",
""
],
[
"Paolini",
"Gianluca",
""
]
] |
2407.02268 | Tanujit Chakraborty | Payel Sadhukhan, Tanujit Chakraborty | Footprints of Data in a Classifier: Understanding the Privacy Risks and
Solution Strategies | null | null | null | null | cs.CR cs.AI | http://creativecommons.org/licenses/by/4.0/ | The widespread deployment of Artificial Intelligence (AI) across government
and private industries brings both advancements and heightened privacy and
security concerns. Article 17 of the General Data Protection Regulation (GDPR)
mandates the Right to Erasure, requiring data to be permanently removed from a
system to prevent potential compromise. While existing research primarily
focuses on erasing sensitive data attributes, several passive data compromise
mechanisms remain underexplored and unaddressed. One such issue arises from the
residual footprints of training data embedded within predictive models.
Performance disparities between test and training data can inadvertently reveal
which data points were part of the training set, posing a privacy risk. This
study examines how two fundamental aspects of classifier systems - training
data quality and classifier training methodology - contribute to privacy
vulnerabilities. Our theoretical analysis demonstrates that classifiers exhibit
universal vulnerability under conditions of data imbalance and distributional
shifts. Empirical findings reinforce our theoretical results, highlighting the
significant role of training data quality in classifier susceptibility.
Additionally, our study reveals that a classifier's operational mechanism and
architectural design impact its vulnerability. We further investigate
mitigation strategies through data obfuscation techniques and analyze their
impact on both privacy and classification performance. To aid practitioners, we
introduce a privacy-performance trade-off index, providing a structured
approach to balancing privacy protection with model effectiveness. The findings
offer valuable insights for selecting classifiers and curating training data in
diverse real-world applications.
| [
{
"version": "v1",
"created": "Tue, 2 Jul 2024 13:56:37 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 08:36:16 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Sadhukhan",
"Payel",
""
],
[
"Chakraborty",
"Tanujit",
""
]
] |
2407.02472 | Hayoung Jung | Chan Young Park, Shuyue Stella Li, Hayoung Jung, Svitlana Volkova,
Tanushree Mitra, David Jurgens, Yulia Tsvetkov | ValueScope: Unveiling Implicit Norms and Values via Return Potential
Model of Social Interactions | First three authors contributed equally. Accepted at EMNLP Findings
2024 | null | 10.18653/v1/2024.findings-emnlp.972 | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | This study introduces ValueScope, a framework leveraging language models to
quantify social norms and values within online communities, grounded in social
science perspectives on normative structures. We employ ValueScope to dissect
and analyze linguistic and stylistic expressions across 13 Reddit communities
categorized under gender, politics, science, and finance. Our analysis provides
a quantitative foundation showing that even closely related communities exhibit
remarkably diverse norms. This diversity supports existing theories and adds a
new dimension--community preference--to understanding community interactions.
ValueScope not only delineates differing social norms among communities but
also effectively traces their evolution and the influence of significant
external events like the U.S. presidential elections and the emergence of new
sub-communities. The framework thus highlights the pivotal role of social norms
in shaping online interactions, presenting a substantial advance in both the
theory and application of social norm studies in digital spaces.
| [
{
"version": "v1",
"created": "Tue, 2 Jul 2024 17:51:27 GMT"
},
{
"version": "v2",
"created": "Mon, 7 Oct 2024 18:04:06 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Park",
"Chan Young",
""
],
[
"Li",
"Shuyue Stella",
""
],
[
"Jung",
"Hayoung",
""
],
[
"Volkova",
"Svitlana",
""
],
[
"Mitra",
"Tanushree",
""
],
[
"Jurgens",
"David",
""
],
[
"Tsvetkov",
"Yulia",
""
]
] |
2407.02636 | Alexander Marin\v{s}ek | Alexander Marinsek, Sam De Kunst, Gilles Callebaut, Lieven De
Strycker, Liesbet Van der Perre | MmWave for Extended Reality: Open User Mobility Dataset,
Characterisation, and Impact on Link Quality | In the process of being published in the IEEE Communications
Magazine, special issue FT2304 / eXtended Reality | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | User mobility in extended reality (XR) can have a major impact on
millimeter-wave (mmWave) links and may require dedicated mitigation strategies
to ensure reliable connections and avoid outage. The available prior art has
predominantly focused on XR applications with constrained user mobility and
limited impact on mmWave channels. We have performed dedicated experiments to
extend the characterisation of relevant future XR use cases featuring a high
degree of user mobility. To this end, we have carried out a tailor-made
measurement campaign and conducted a characterisation of the collected tracking
data, including the approximation of the data using statistical distributions.
Moreover, we have provided an interpretation of the possible impact of the
recorded mobility on mmWave technology. The dataset is made publicly accessible
to provide a testing ground for wireless system design and to enable further XR
mobility modelling.
| [
{
"version": "v1",
"created": "Tue, 2 Jul 2024 20:05:17 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 07:39:09 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Marinsek",
"Alexander",
""
],
[
"De Kunst",
"Sam",
""
],
[
"Callebaut",
"Gilles",
""
],
[
"De Strycker",
"Lieven",
""
],
[
"Van der Perre",
"Liesbet",
""
]
] |
2407.03602 | Michael Sadovskii | M.V. Sadovskii | Upper limit for superconducting transition temperature in
Eliashberg-McMillan theory | 2 pages, minor typos corrected | JETP Letters v. 120, No.3, 205-207 (2024) | 10.1134/S0021364024602409 | null | cond-mat.supr-con | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present simple qualitative estimates for the maximal superconducting
transition temperature, which may be achieved due to electron - phonon coupling
in Eliashberg-McMillan theory. It is shown that in the limit of very strong
coupling the upper limit for transition temperature is determined in fact by a
combination of atomic constants and density of conduction electrons.
| [
{
"version": "v1",
"created": "Thu, 4 Jul 2024 03:26:46 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 05:54:21 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Sadovskii",
"M. V.",
""
]
] |
2407.04082 | Saurabhchand Bhati | Saurabhchand Bhati, Yuan Gong, Leonid Karlinsky, Hilde Kuehne, Rogerio
Feris, James Glass | DASS: Distilled Audio State Space Models Are Stronger and More
Duration-Scalable Learners | null | null | null | null | eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | State-space models (SSMs) have emerged as an alternative to Transformers for
audio modeling due to their high computational efficiency with long inputs.
While recent efforts on Audio SSMs have reported encouraging results, two main
limitations remain: First, in 10-second short audio tagging tasks, Audio SSMs
still underperform compared to Transformer-based models such as Audio
Spectrogram Transformer (AST). Second, although Audio SSMs theoretically
support long audio inputs, their actual performance with long audio has not
been thoroughly evaluated. To address these limitations, in this paper, 1) We
applied knowledge distillation in audio space model training, resulting in a
model called Knowledge Distilled Audio SSM (DASS). To the best of our
knowledge, it is the first SSM that outperforms the Transformers on AudioSet
and achieves an mAP of 48.9; and 2) We designed a new test called Audio Needle
In A Haystack (Audio NIAH). We find that DASS, trained with only 10-second
audio clips, can retrieve sound events in audio recordings up to 2.5 hours
long, while the AST model fails when the input is just 50 seconds,
demonstrating SSMs are indeed more duration scalable. Code:
https://github.com/Saurabhbhati/DASS,
https://huggingface.co/saurabhati/DASS_small_AudioSet_48.9
| [
{
"version": "v1",
"created": "Thu, 4 Jul 2024 17:46:19 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 19:25:38 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Bhati",
"Saurabhchand",
""
],
[
"Gong",
"Yuan",
""
],
[
"Karlinsky",
"Leonid",
""
],
[
"Kuehne",
"Hilde",
""
],
[
"Feris",
"Rogerio",
""
],
[
"Glass",
"James",
""
]
] |
2407.04380 | Rafael Sayous | Rafael Sayous | Gaps in the complex Farey sequence of an imaginary quadratic number
field | 20 pages, 6 figures | null | null | null | math.NT math.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Given an imaginary quadratic number field $K$ with ring of integers
$\mathcal{O}_K$, we are interested in the asymptotic \emph{distance to nearest
neighbour} (or \emph{gap}) statistic of complex Farey fractions $\frac{p}{q}$,
with $p,q \in \mathcal{O}_K$ and $0<|q|\leq T$, as $T \to \infty$.
Reformulating this problem in a homogeneous dynamical setting, we follow the
approach of J. Marklof for real Farey fractions with several variables (2013)
and adapt a joint equidistribution result in the real $3$-dimensional
hyperbolic space of J. Parkkonen and F. Paulin (2023) to derive the existence
of a probability measure describing this asymptotic gap statistic. We obtain an
integral formula for the associated cumulative distribution function, and use
geometric arguments to find an explicit estimate for its tail distribution in
the cases of Gaussian and Eisenstein fractions.
| [
{
"version": "v1",
"created": "Fri, 5 Jul 2024 09:33:46 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 09:51:49 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Sayous",
"Rafael",
""
]
] |
2407.04456 | You-Wei Benson Chen | You-Wei Benson Chen and Alejandro Claros | $\beta$-dimensional sharp maximal function and applications | 30 pages | null | null | null | math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we study $\beta$-dimensional sharp maximal operator defined as
\begin{align*} \mathcal{M}^{\#} _\beta f(x) := \sup_{Q} \inf_{c \in \mathbb{R}}
\chi_{Q}(x) \frac{1}{\ell(Q)^\beta} \int_Q |f-c| \; d
\mathcal{H}^{\beta}_\infty, \end{align*} where the supremum is taken over all
cubes in $\mathbb{R}^d$ with sides pararell to the coordinate axes, $\ell(Q)$
is the length side of $Q$ and $\mathcal{H}^{\beta}_\infty$ is the Hausdorff
content. In particular, we prove Fefferman-Stein inequality for
$\mathcal{M}^{\#} _\beta f$ by giving a good lambda estimate for
$\beta$-dimensional sharp maximal operator in the context of Hausdorff content.
Additionally, we prove the Muckenhoupt-Wheeden inequality in this framework by
establishing a good lambda inequality of independent interest.
| [
{
"version": "v1",
"created": "Fri, 5 Jul 2024 12:08:01 GMT"
},
{
"version": "v2",
"created": "Tue, 4 Mar 2025 10:44:44 GMT"
},
{
"version": "v3",
"created": "Sat, 12 Apr 2025 18:29:49 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Chen",
"You-Wei Benson",
""
],
[
"Claros",
"Alejandro",
""
]
] |
2407.04492 | Dingyuan Liu | Dingyuan Liu, Let\'icia Mattos, Tibor Szab\'o | On the number of sets with small sumset | 30 pages + appendix | null | null | null | math.CO math.NT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate subsets with small sumset in arbitrary abelian groups. For an
abelian group $G$ and an $n$-element subset $Y \subseteq G$ we show that if $m
\ll s^2/(\log n)^2$, then the number of subsets $A \subseteq Y$ with $|A| = s$
and $|A + A| \leq m$ is at most \[2^{o(s)}\binom{\frac{m+\beta}{2}}{s},\] where
$\beta$ is the size of the largest subgroup of $G$ of size at most
$\left(1+o(1)\right)m$. This bound is sharp for $\mathbb{Z}$ and many other
groups. Our result improves the one of Campos and nearly bridges the remaining
gap in a conjecture of Alon, Balogh, Morris, and Samotij.
We also explore the behaviour of uniformly chosen random sets $A \subseteq
\{1,\ldots,n\}$ with $|A| = s$ and $|A + A| \leq m$. Under the same assumption
that $m \ll s^2/(\log n)^2$, we show that with high probability there exists an
arithmetic progression $P \subseteq \mathbb{Z}$ of size at most $m/2 + o(m)$
containing all but $o(s)$ elements of $A$. Analogous results are obtained for
asymmetric sumsets, improving results by Campos, Coulson, Serra, and W\"otzel.
The main tool behind our results is a more efficient container-type theorem
developed for sets with small sumset, which gives an essentially optimal
collection of containers. The proof of this combines an adapted hypergraph
container lemma, that caters to the asymmetric setup as well, with a novel
``preprocessing'' graph container lemma, which allows the hypergraph container
lemma to be called upon significantly less times than was necessary before.
| [
{
"version": "v1",
"created": "Fri, 5 Jul 2024 13:30:27 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 12:30:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Liu",
"Dingyuan",
""
],
[
"Mattos",
"Letícia",
""
],
[
"Szabó",
"Tibor",
""
]
] |
2407.04660 | Thomas Bakx `` | Thomas Bakx, Nora Elisa Chisari and Zvonimir Vlah | COBRA: Optimal Factorization of Cosmological Observables | 5+6 pages, substantial rewriting, conclusions unchanged. PRL accepted
version | null | null | null | astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | We introduce COBRA (Cosmology with Optimally factorized Bases of Radial
Approximants), a novel framework for rapid computation of large-scale structure
observables. COBRA separates scale dependence from cosmological parameters in
the linear matter power spectrum while also minimising the number of necessary
basis terms $N_b$, thus enabling direct and efficient computation of derived
and nonlinear observables. Moreover, the dependence on cosmological parameters
is efficiently approximated using radial basis function interpolation. We apply
our framework to decompose the linear matter power spectrum in the standard
$\Lambda$CDM scenario, as well as by adding curvature, dynamical dark energy
and massive neutrinos, covering all redshifts relevant for Stage IV surveys.
With only a dozen basis terms $N_b$, COBRA reproduces exact Boltzmann solver
calculations to $\sim 0.1\%$ precision, which improves further to $0.02\%$ in
the pure $\Lambda$CDM scenario. Using our decomposition, we recast the one-loop
redshift space galaxy power spectrum in a separable minimal-basis form,
enabling $\sim 4000$ model evaluations per second at $0.02\%$ precision on a
single thread. This constitutes a considerable improvement over previously
existing methods (e.g., FFTLog) opening a window for efficient computations of
higher loop and higher order correlators involving multiple powers of the
linear matter power spectra. The resulting factorisation can also be utilised
in clustering, weak lensing and CMB analyses. Our implementation will be made
public upon publication.
| [
{
"version": "v1",
"created": "Fri, 5 Jul 2024 17:17:38 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 09:39:39 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Bakx",
"Thomas",
""
],
[
"Chisari",
"Nora Elisa",
""
],
[
"Vlah",
"Zvonimir",
""
]
] |
2407.04975 | Gideon Koekoek | Richard Kerner, Gideon Koekoek, Julia Schuring, Jan-Willem van Holten | Polar magnetic fields in black-hole space-times | null | JCAP 03 (2025) 065 | null | null | gr-qc | http://creativecommons.org/licenses/by-nc-nd/4.0/ | To model magnetic fields of compact objects we solve the Maxwell equations in
the background of the exterior static Schwarzschild and slowly rotating Kerr
space-times. We impose the boundary condition that the electromagnetic fields
are to vanish at infinity. A full set of solutions is obtained, describing
axially symmetric magnetic fields, supplemented by axial electric fields in the
case of non-vanishing rotation of the gravitational background. We study the
motion of charged test particles in these combined gravitational and
electromagnetic fields, in particular considering the conditions for circular
equatorial orbits. Such orbits always exist in odd-multipole magnetic fields,
and they can exist for particular radii in a combination of two or more
even-multipole magnetic fields. Combinations of several odd-multipole fields
can give rise to radial variation in the field orientation and the direction of
motion of charged particles. Deviations from circularity are described using a
perturbative approach. This also allows to study the stability of the parent
circular orbits.
| [
{
"version": "v1",
"created": "Sat, 6 Jul 2024 06:55:52 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Kerner",
"Richard",
""
],
[
"Koekoek",
"Gideon",
""
],
[
"Schuring",
"Julia",
""
],
[
"van Holten",
"Jan-Willem",
""
]
] |
2407.06165 | Moritz Rempe | M. Rempe, F. H\"orst, C. Seibold, B. Hadaschik, M. Schlimbach, J.
Egger, K. Kr\"oninger, F. Breuer, M. Blaimer, J. Kleesiek | Tumor likelihood estimation on MRI prostate data by utilizing k-Space
information | null | null | null | null | cs.CV cs.AI physics.med-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We present a novel preprocessing and prediction pipeline for the
classification of magnetic resonance imaging (MRI) that takes advantage of the
information rich complex valued k-Space. Using a publicly available MRI raw
dataset with 312 subject and a total of 9508 slices, we show the advantage of
utilizing the k-Space for better prostate cancer likelihood estimation in
comparison to just using the magnitudinal information in the image domain, with
an AUROC of $86.1\%\pm1.8\%$. Additionally, by using high undersampling rates
and a simple principal component analysis (PCA) for coil compression, we reduce
the time needed for reconstruction by avoiding the time intensive GRAPPA
reconstruction algorithm. By using digital undersampling for our experiments,
we show that scanning and reconstruction time could be reduced. Even with an
undersampling factor of 16, our approach achieves meaningful results, with an
AUROC of $71.4\%\pm2.9\%$, using the PCA coil combination and taking into
account the k-Space information. With this study, we were able to show the
feasibility of preserving phase and k-Space information, with consistent
results. Besides preserving valuable information for further diagnostics, this
approach can work without the time intensive ADC and reconstruction
calculations, greatly reducing the post processing, as well as potential
scanning time, increasing patient comfort and allowing a close to real-time
prediction.
| [
{
"version": "v1",
"created": "Tue, 4 Jun 2024 12:05:20 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 10:28:26 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Rempe",
"M.",
""
],
[
"Hörst",
"F.",
""
],
[
"Seibold",
"C.",
""
],
[
"Hadaschik",
"B.",
""
],
[
"Schlimbach",
"M.",
""
],
[
"Egger",
"J.",
""
],
[
"Kröninger",
"K.",
""
],
[
"Breuer",
"F.",
""
],
[
"Blaimer",
"M.",
""
],
[
"Kleesiek",
"J.",
""
]
] |
2407.07407 | Takafumi Miyazaki | Maohua Le and Takafumi Miyazaki | General sharp bounds for the number of solutions to purely exponential
equations with three terms | 25 pages; title changed; major revision | null | null | null | math.NT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | It is conjectured that for any fixed relatively prime positive integers $a,b$
and $c$ all greater than 1 there is at most one solution to the equation
$a^x+b^y=c^z$ in positive integers $x,y$ and $z$, except for specific cases. In
this paper, we prove that for any fixed $c$ there is at most one solution to
the equation, except for only finitely many pairs of $a$ and $b.$ This is
regarded as a 3-variable generalization of the result of Miyazaki and Pink [T.
Miyazaki and I. Pink, Number of solutions to a special type of unit equations
in two unknowns, III, arXiv:2403.20037 (accepted for publication in Math. Proc.
Cambridge Philos. Soc.)] which asserts that for any fixed positive integer $a$
there are only finitely many pairs of coprime positive integers $b$ and $c$
with $b>1$ such that the Pillai's type equation $a^x-b^y=c$ has more than one
solution in positive integers $x$ and $y$. The proof of our result is based on
a certain $p$-adic idea of Miyazaki and Pink and relies on many deep theorems
on the theory of Diophantine approximation, and it also includes the complete
description of solutions to some interesting system of simultaneous
polynomial-exponential equations. We also discuss how effectively exceptional
pairs of $a$ and $b$ on our result for each $c$ can be determined.
| [
{
"version": "v1",
"created": "Wed, 10 Jul 2024 07:08:08 GMT"
},
{
"version": "v2",
"created": "Wed, 17 Jul 2024 22:58:52 GMT"
},
{
"version": "v3",
"created": "Sun, 13 Apr 2025 05:00:47 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Le",
"Maohua",
""
],
[
"Miyazaki",
"Takafumi",
""
]
] |
2407.07603 | Yunusa Haruna | Haruna Yunusa, Qin Shiyin, Abdulrahman Hamman Adama Chukkol, Adamu
Lawan, Abdulganiyu Abdu Yusuf, Isah Bello | iiANET: Inception Inspired Attention Hybrid Network for efficient
Long-Range Dependency | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The recent emergence of hybrid models has introduced a transformative
approach to computer vision, gradually moving beyond conventional convolutional
neural net-works and vision transformers. However, efficiently combining these
two paradigms to better capture long-range dependencies in complex images
remains a challenge. In this paper, we present iiANET (Inception Inspired
Attention Network), an efficient hybrid visual backbone designed to improve the
modeling of long-range dependen-cies. The core innovation of iiANET is the
iiABlock, a unified building block that in-tegrates global r-MHSA (Multi-Head
Self-Attention) and convolutional layers in paral-lel. This design enables
iiABlock to simultaneously capture global context and local details, making it
highly effective for extracting rich and diverse features. By effi-ciently
fusing these complementary representations, iiABlock allows iiANET to achieve
strong feature interaction while maintaining computational efficiency.
Exten-sive qualitative and quantitative evaluations across various benchmarks
show im-proved performance over several state-of-the-art models.
| [
{
"version": "v1",
"created": "Wed, 10 Jul 2024 12:39:02 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 11:32:38 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Yunusa",
"Haruna",
""
],
[
"Shiyin",
"Qin",
""
],
[
"Chukkol",
"Abdulrahman Hamman Adama",
""
],
[
"Lawan",
"Adamu",
""
],
[
"Yusuf",
"Abdulganiyu Abdu",
""
],
[
"Bello",
"Isah",
""
]
] |
2407.07703 | Xiaolei Wu | Fan Wu, Xiaolei Wu, Mengfei Zhao, Zixiang Zhou | Embedding groups into boundedly acyclic groups | Added a new section about l2-invisibility, some other small changes.
42pages. Final version, to appear in J. Lond. Math. Soc | null | null | null | math.GR math.KT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that the \s{\phi}-labeled Thompson groups and the twisted
Brin--Thompson groups are boundedly acyclic. This allows us to prove several
new embedding results for groups. First, every group of type $F_n$ embeds
quasi-isometrically into a boundedly acyclic group of type $F_n$ that has no
proper finite index subgroups. This improves a result of Bridson and a theorem
of Fournier-Facio--L\"oh--Moraschini. Second, every group of type $F_n$ embeds
quasi-isometrically into a $5$-uniformly perfect group of type $F_n$. Third,
using Belk--Zaremsky's construction of twisted Brin--Thompson groups, we show
that every finitely generated group embeds quasi-isometrically into a finitely
generated boundedly acyclic simple group. We also partially answer some
questions of Brothier and Tanushevski regarding the finiteness property of
$\phi$-labeled Thompson group $V_\phi(G)$ and $F_\phi(G)$.
| [
{
"version": "v1",
"created": "Wed, 10 Jul 2024 14:34:45 GMT"
},
{
"version": "v2",
"created": "Tue, 6 Aug 2024 07:00:16 GMT"
},
{
"version": "v3",
"created": "Fri, 28 Feb 2025 07:32:06 GMT"
},
{
"version": "v4",
"created": "Sun, 13 Apr 2025 08:25:28 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Wu",
"Fan",
""
],
[
"Wu",
"Xiaolei",
""
],
[
"Zhao",
"Mengfei",
""
],
[
"Zhou",
"Zixiang",
""
]
] |
2407.08388 | Geoff Keeling | Geoff Keeling, Winnie Street | On the attribution of confidence to large language models | 22 pages, 0 figures | null | 10.1080/0020174X.2025.2450598 | null | cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | Credences are mental states corresponding to degrees of confidence in
propositions. Attribution of credences to Large Language Models (LLMs) is
commonplace in the empirical literature on LLM evaluation. Yet the theoretical
basis for LLM credence attribution is unclear. We defend three claims. First,
our semantic claim is that LLM credence attributions are (at least in general)
correctly interpreted literally, as expressing truth-apt beliefs on the part of
scientists that purport to describe facts about LLM credences. Second, our
metaphysical claim is that the existence of LLM credences is at least
plausible, although current evidence is inconclusive. Third, our epistemic
claim is that LLM credence attributions made in the empirical literature on LLM
evaluation are subject to non-trivial sceptical concerns. It is a distinct
possibility that even if LLMs have credences, LLM credence attributions are
generally false because the experimental techniques used to assess LLM
credences are not truth-tracking.
| [
{
"version": "v1",
"created": "Thu, 11 Jul 2024 10:51:06 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Keeling",
"Geoff",
""
],
[
"Street",
"Winnie",
""
]
] |
2407.08576 | Pierre-Fran\c{c}ois Loos Dr | Yann Damour and Anthony Scemama and F\'abris Kossoski and
Pierre-Fran\c{c}ois Loos | Selected Configuration Interaction for Resonances | 10 pages, 2 figures (supporting information available) | J. Phys. Chem. Lett. 15, 8296 (2024) | 10.1021/acs.jpclett.4c02060 | null | physics.chem-ph cond-mat.mtrl-sci nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Electronic resonances are metastable states that can decay by electron loss.
They are ubiquitous across various fields of science, such as chemistry,
physics, and biology. However, current theoretical and computational models for
resonances cannot yet rival the level of accuracy achieved by bound-state
methodologies. Here, we generalize selected configuration interaction (SCI) to
treat resonances using the complex absorbing potential (CAP) technique. By
modifying the selection procedure and the extrapolation protocol of standard
SCI, the resulting CAP-SCI method yields resonance positions and widths of full
configuration interaction quality. Initial results for the shape resonances of
\ce{N2-} and \ce{CO-} reveal the important effect of high-order correlation,
which shifts the values obtained with CAP-augmented equation-of-motion
coupled-cluster with singles and doubles by more than \SI{0.1}{\eV}. The
present CAP-SCI approach represents a cornerstone in the development of
highly-accurate methodologies for resonances.
| [
{
"version": "v1",
"created": "Thu, 11 Jul 2024 15:02:06 GMT"
},
{
"version": "v2",
"created": "Thu, 8 Aug 2024 15:21:49 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Damour",
"Yann",
""
],
[
"Scemama",
"Anthony",
""
],
[
"Kossoski",
"Fábris",
""
],
[
"Loos",
"Pierre-François",
""
]
] |
2407.08804 | Gokul Pathikonda | Isaiah E. Wall, Gokul Pathikonda | Coherent Organization of Passive Scalar from a Point-Source in a
Turbulent Boundary Layer | null | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | The spatial organization of a passive scalar plume originating from a point
source in a turbulent boundary layer is studied to understand its meandering
characteristics. We focus shortly downstream of the isokinetic injection
($1.5\le x/\delta \le 3$, $\delta$ being boundary layer thickness) where the
scalar concentration is highly intermittent, the plume rapidly
\textit{meanders}, and \textit{breaks-up} into concentrated scalar pockets due
to the action of turbulent structures. Two injection locations were considered:
the center of logarithmic-region and the wake-region of the boundary layer.
Simultaneous quantitative acetone planar laser-induced fluorescence (Ac-PLIF)
and particle-image velocimetry (PIV) were performed in a wind-tunnel, to
measure scalar mixture fraction and velocity fields. Single- and multi-point
statistics were compared to established works to validate the diagnostic
novelties. Additionally, the spatial characteristics of plume intermittency
were quantified using `blob' size, shape, orientation and mean concentration.
It was observed that straining, break up and spatial reorganization were the
primary plume-evolution modes in this region, with little small-scale
homogenization. Further, the dominant role of coherent vortex motions in plume
meandering and break-up was evident. Their action is found to be the primary
mechanism by which the injected scalar is transported away from the wall in
high concentrations (`large meander events (LMEs)'). Strong spatial correlation
was observed in both instantaneous and conditional fields between the high
concentration regions and individual vortex heads. This coherent transport was
weaker for wake-injection, where the plume only interacts with outer vortex
motions. A coherent-structure based mechanism is suggested to explain these
transport mechanisms.
| [
{
"version": "v1",
"created": "Thu, 11 Jul 2024 18:24:24 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 00:05:37 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Wall",
"Isaiah E.",
""
],
[
"Pathikonda",
"Gokul",
""
]
] |
2407.09060 | Pierre-Fran\c{c}ois Loos Dr | Iryna Knysh and Filippo Lipparini and Aymeric Blondel and Ivan
Duchemin and Xavier Blase and Pierre-Fran\c{c}ois Loos and Denis Jacquemin | Reference CC3 Excitation Energies for Organic Chromophores: Benchmarking
TD-DFT, BSE/$GW$ and Wave Function Methods | 27 pages, 10 figures (Supporting Information available) | J. Chem. Theory Comput. 20, 8152 (2024) | 10.1021/acs.jctc.4c00906 | null | physics.chem-ph cond-mat.mtrl-sci cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | To expand the QUEST database of highly-accurate vertical transition energies,
we consider a series of large organic chromogens ubiquitous in dye chemistry,
such as anthraquinone, azobenzene, BODIPY, and naphthalimide. We compute, at
the CC3 level of theory, the singlet and triplet vertical transition energies
associated with the low-lying excited states. This leads to a collection of
more than 120 new highly-accurate excitation energies. Subsequently, we employ
these reference values to benchmark a series of lower-order wave function
approaches, including the popular ADC(2) and CC2 schemes, as well as
time-dependent density-functional theory (TD-DFT), both with and without
applying the Tamm-Dancoff approximation (TDA). At the TD-DFT level, we evaluate
a large panel of global, range-separated, local, and double hybrid functionals.
Additionally, we assess the performance of the Bethe-Salpeter equation (BSE)
formalism relying on both $G_0W_0$ and ev$GW$ quasiparticle energies evaluated
from various starting points. It turns out that CC2 and ADC(2.5) are the most
accurate models amongst those with respective $\mathcal{O}(N^5)$ and
$\mathcal{O}(N^6)$ scalings with system size. In contrast, CCSD does not
outperform CC2. The best performing exchange-correlation functionals include
BMK, M06-2X, M06-SX, CAM-B3LYP, $\omega$B97X-D, and LH20t, with average
deviations of approximately 0.20 eV or slightly below. Errors on vertical
excitation energies can be further reduced by considering double hybrids. Both
SOS-$\omega$B88PP86 and SOS-$\omega$PBEPP86 exhibit particularly attractive
performances with overall quality on par with CC2, whereas PBE0-DH and PBE-QIDH
are only slightly less efficient. BSE/ev$GW$ calculations based on Kohn-Sham
starting points have been found to be particularly effective for singlet
transitions, but much less for their triplet counterparts.
| [
{
"version": "v1",
"created": "Fri, 12 Jul 2024 07:28:27 GMT"
},
{
"version": "v2",
"created": "Mon, 26 Aug 2024 13:08:15 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Knysh",
"Iryna",
""
],
[
"Lipparini",
"Filippo",
""
],
[
"Blondel",
"Aymeric",
""
],
[
"Duchemin",
"Ivan",
""
],
[
"Blase",
"Xavier",
""
],
[
"Loos",
"Pierre-François",
""
],
[
"Jacquemin",
"Denis",
""
]
] |
2407.09647 | Timothy J. Pearson | S. Kiehlmann, P. V. de la Parra, A. G. Sullivan, A. Synani, I.
Liodakis, P. Mr\'oz, S. K. N{\ae}ss, A. C. S. Readhead, M. C. Begelman, R. D.
Blandford, K. Chatziioannou, Y. Ding, M. J. Graham, F. Harrison, D. C. Homan,
T. Hovatta, S. R. Kulkarni, M. L. Lister, R. Maiolino, W. Max-Moerbeck, B.
Molina, C. P. O'Dea, V. Pavlidou, T. J. Pearson, M. F. Aller, C. R. Lawrence,
T. J. W. Lazio, S. O'Neill, T. A. Prince, V. Ravi, R. A. Reeves, K. Tassis,
M. Vallisneri, J. A. Zensus | PKS 2131-021 -- Discovery of Strong Coherent Sinusoidal Variations from
Radio to Optical Frequencies: Compelling Evidence for a Blazar Supermassive
Black Hole Binary | 30 pages, 20 figures, 5 tables. Accepted for publication in ApJ | null | null | null | astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | Haystack and Owens Valley Radio Observatory (OVRO) observations recently
revealed strong, intermittent, sinusoidal total flux-density variations that
maintained coherence between 1975 and 2021 in the blazar PKS 2131$-$021
($z=1.283$). This was interpreted as possible evidence of a supermassive black
hole binary (SMBHB). Extended observations through 2023 show coherence over
47.9 years, with an observed period $P_\textrm{15 GHz}=(1739.8 \pm 17.4)$
days}. We reject, with $p$-value = $2.09 \times 10^{-7}$, the hypothesis that
the variations are due to random fluctuations in the red noise tail of the
power spectral density. There is clearly a physical phenomenon in PKS
2131$-$021 producing coherent sinusoidal flux density variations. We find the
coherent sinusoidal intensity variations extend from below 2.7 GHz to optical
frequencies, from which we derive an observed period $P_\textrm{optical}=(1764
\pm 36)$ days. Across this broad frequency range there is a smoothly-varying
monotonic phase shift in the sinusoidal variations with frequency. Hints of
periodic variations are also observed at $\gamma$-ray energies. The importance
of well-vetted SMBHB candidates to searches for gravitational waves is pointed
out. We estimate the fraction of blazars that are SMBHB candidates to be > 1 in
100. Thus monitoring programs covering tens of thousands of blazars could
discover hundreds of SMBHB candidates.
| [
{
"version": "v1",
"created": "Fri, 12 Jul 2024 19:05:24 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Apr 2025 20:51:18 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Kiehlmann",
"S.",
""
],
[
"de la Parra",
"P. V.",
""
],
[
"Sullivan",
"A. G.",
""
],
[
"Synani",
"A.",
""
],
[
"Liodakis",
"I.",
""
],
[
"Mróz",
"P.",
""
],
[
"Næss",
"S. K.",
""
],
[
"Readhead",
"A. C. S.",
""
],
[
"Begelman",
"M. C.",
""
],
[
"Blandford",
"R. D.",
""
],
[
"Chatziioannou",
"K.",
""
],
[
"Ding",
"Y.",
""
],
[
"Graham",
"M. J.",
""
],
[
"Harrison",
"F.",
""
],
[
"Homan",
"D. C.",
""
],
[
"Hovatta",
"T.",
""
],
[
"Kulkarni",
"S. R.",
""
],
[
"Lister",
"M. L.",
""
],
[
"Maiolino",
"R.",
""
],
[
"Max-Moerbeck",
"W.",
""
],
[
"Molina",
"B.",
""
],
[
"O'Dea",
"C. P.",
""
],
[
"Pavlidou",
"V.",
""
],
[
"Pearson",
"T. J.",
""
],
[
"Aller",
"M. F.",
""
],
[
"Lawrence",
"C. R.",
""
],
[
"Lazio",
"T. J. W.",
""
],
[
"O'Neill",
"S.",
""
],
[
"Prince",
"T. A.",
""
],
[
"Ravi",
"V.",
""
],
[
"Reeves",
"R. A.",
""
],
[
"Tassis",
"K.",
""
],
[
"Vallisneri",
"M.",
""
],
[
"Zensus",
"J. A.",
""
]
] |
2407.10989 | Sandra C Matz | Sandra C. Matz, Heinrich Peters, Moran Cerf, Eric Grunenberg, Paul W.
Eastwick, Mitja D. Back and Eli J. Finkel | Can Large Language Models Detect Verbal Indicators of Romantic
Attraction? | null | null | null | null | cs.CL cs.AI cs.HC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | As artificial intelligence (AI) models become an integral part of everyday
life, our interactions with them shift from purely functional exchanges to more
relational experiences. For these experiences to be successful, artificial
agents need to be able to detect and interpret social cues and interpersonal
dynamics; both within and outside of their own human-agent relationships. In
this paper, we explore whether AI models can accurately decode one of the
arguably most important but complex social signals: romantic attraction.
Specifically, we test whether Large Language Models can detect romantic
attraction during brief getting-to-know-you interactions between humans.
Examining data from 964 speed dates, we show that ChatGPT can predict both
objective and subjective indicators of speed dating success (r=0.12-0.23).
Although predictive performance remains relatively low, ChatGPT's predictions
of actual matching (i.e., the exchange of contact information) were not only on
par with those of human judges but incremental to speed daters' own
predictions. In addition, ChatGPT's judgments showed substantial overlap with
those made by human observers (r=0.21-0.35), highlighting similarities in their
representation of romantic attraction that are independent of accuracy. Our
findings also offer insights into how ChatGPT arrives at its predictions and
the mistakes it makes. Specifically, we use a Brunswik lens approach to
identify the linguistic and conversational cues utilized by ChatGPT (and human
judges) vis-a-vis those that are predictive of actual matching.
| [
{
"version": "v1",
"created": "Sun, 23 Jun 2024 17:50:30 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 20:24:02 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Matz",
"Sandra C.",
""
],
[
"Peters",
"Heinrich",
""
],
[
"Cerf",
"Moran",
""
],
[
"Grunenberg",
"Eric",
""
],
[
"Eastwick",
"Paul W.",
""
],
[
"Back",
"Mitja D.",
""
],
[
"Finkel",
"Eli J.",
""
]
] |
2407.11211 | Philipp Allgeuer | Philipp Allgeuer and Kyra Ahrens and Stefan Wermter | Unconstrained Open Vocabulary Image Classification: Zero-Shot Transfer
from Text to Image via CLIP Inversion | Published at WACV 2025 | Winter Conference on Applications of Computer Vision (WACV), 2025,
pp. 8206-8217 | 10.1109/WACV61041.2025.00797 | null | cs.CV cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | We introduce NOVIC, an innovative real-time uNconstrained Open Vocabulary
Image Classifier that uses an autoregressive transformer to generatively output
classification labels as language. Leveraging the extensive knowledge of CLIP
models, NOVIC harnesses the embedding space to enable zero-shot transfer from
pure text to images. Traditional CLIP models, despite their ability for open
vocabulary classification, require an exhaustive prompt of potential class
labels, restricting their application to images of known content or context. To
address this, we propose an "object decoder" model that is trained on a
large-scale 92M-target dataset of templated object noun sets and LLM-generated
captions to always output the object noun in question. This effectively inverts
the CLIP text encoder and allows textual object labels from essentially the
entire English language to be generated directly from image-derived embedding
vectors, without requiring any a priori knowledge of the potential content of
an image, and without any label biases. The trained decoders are tested on a
mix of manually and web-curated datasets, as well as standard image
classification benchmarks, and achieve fine-grained prompt-free prediction
scores of up to 87.5%, a strong result considering the model must work for any
conceivable image and without any contextual clues.
| [
{
"version": "v1",
"created": "Mon, 15 Jul 2024 19:53:02 GMT"
},
{
"version": "v2",
"created": "Wed, 17 Jul 2024 22:23:42 GMT"
},
{
"version": "v3",
"created": "Mon, 18 Nov 2024 14:43:38 GMT"
},
{
"version": "v4",
"created": "Tue, 26 Nov 2024 09:28:35 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Allgeuer",
"Philipp",
""
],
[
"Ahrens",
"Kyra",
""
],
[
"Wermter",
"Stefan",
""
]
] |
2407.11218 | Renchi Zhang | Renchi Zhang, Jesse van der Linden, Dimitra Dodou, Harleigh Seyffert,
Yke Bauke Eisma, Joost C. F. de Winter | Walk along: An Experiment on Controlling the Mobile Robot 'Spot' with
Voice and Gestures | null | null | null | null | cs.HC cs.RO | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Robots are becoming more capable and can autonomously perform tasks such as
navigating between locations. However, human oversight remains crucial. This
study compared two touchless methods for directing mobile robots: voice control
and gesture control, to investigate the efficiency of the methods and the
preference of users. We tested these methods in two conditions: one in which
participants remained stationary and one in which they walked freely alongside
the robot. We hypothesized that walking alongside the robot would result in
higher intuitiveness ratings and improved task performance, based on the idea
that walking promotes spatial alignment and reduces the effort required for
mental rotation. In a 2x2 within-subject design, 218 participants guided the
quadruped robot Spot along a circuitous route with multiple 90-degree turns
using rotate left, rotate right, and walk forward commands. After each trial,
participants rated the intuitiveness of the command mapping, while
post-experiment interviews were used to gather the participants' preferences.
Results showed that voice control combined with walking with Spot was the most
favored and intuitive, whereas gesture control while standing caused confusion
for left/right commands. Nevertheless, 29% of participants preferred gesture
control, citing increased task engagement and visual congruence as reasons. An
odometry-based analysis revealed that participants often followed behind Spot,
particularly in the gesture control condition, when they were allowed to walk.
In conclusion, voice control with walking produced the best outcomes. Improving
physical ergonomics and adjusting gesture types could make gesture control more
effective.
| [
{
"version": "v1",
"created": "Mon, 15 Jul 2024 20:07:33 GMT"
},
{
"version": "v2",
"created": "Wed, 17 Jul 2024 07:48:45 GMT"
},
{
"version": "v3",
"created": "Mon, 13 Jan 2025 09:23:41 GMT"
},
{
"version": "v4",
"created": "Fri, 4 Apr 2025 16:25:04 GMT"
},
{
"version": "v5",
"created": "Mon, 14 Apr 2025 08:01:06 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Zhang",
"Renchi",
""
],
[
"van der Linden",
"Jesse",
""
],
[
"Dodou",
"Dimitra",
""
],
[
"Seyffert",
"Harleigh",
""
],
[
"Eisma",
"Yke Bauke",
""
],
[
"de Winter",
"Joost C. F.",
""
]
] |
2407.12219 | Alexander Clow A.Clow | Alexander Clow and Neil A McKay | Digraph Placement Games | 30 pages, 3 figures, 1 appendix | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | This paper considers a natural ruleset for playing a partisan combinatorial
game on a directed graph, which we call Digraph Placement. Given a digraph $G$
with a not necessarily proper $2$-coloring of $V(G)$, the Digraph Placement
game played on $G$ by the players Left and Right, who play alternately, is
defined as follows. On her turn, Left chooses a blue vertex which is deleted
along with all of its out-neighbours. On his turn Right chooses a red vertex,
which is deleted along with all of its out-neighbours. A player loses if on
their turn they cannot move. We show constructively that Digraph Placement is a
universal partisan ruleset; for all partisan combinatorial games $X$ there
exists a Digraph Placement game, $G$, such that $G = X$. Digraph Placement and
many other games including Nim, Poset Game, Col, Node Kayles, Domineering, and
Arc Kayles are instances of a class of placement games that we call conflict
placement games. We prove that $X$ is a conflict placement game if and only if
it has the same literal form as a Digraph Placement game. A corollary of this
is that deciding the winner of a Digraph Placement game is PSPACE-hard. Next,
for a game value $X$ we prove bounds on the order of a smallest Digraph
Placement game $G$ such that $G = X$.
| [
{
"version": "v1",
"created": "Tue, 16 Jul 2024 23:57:42 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 15:38:52 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Clow",
"Alexander",
""
],
[
"McKay",
"Neil A",
""
]
] |
2407.12223 | Chengzhi Lin | Chengzhi Lin, Shuchang Liu, Chuyuan Wang, Yongqi Liu | Conditional Quantile Estimation for Uncertain Watch Time in Short-Video
Recommendation | 10 pages, 5 figures, 7 tables | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Accurately predicting watch time is crucial for optimizing recommendations
and user experience in short video platforms. However, existing methods that
estimate a single average watch time often fail to capture the inherent
uncertainty in user engagement patterns. In this paper, we propose Conditional
Quantile Estimation (CQE) to model the entire conditional distribution of watch
time. Using quantile regression, CQE characterizes the complex watch-time
distribution for each user-video pair, providing a flexible and comprehensive
approach to understanding user behavior. We further design multiple strategies
to combine the quantile estimates, adapting to different recommendation
scenarios and user preferences. Extensive offline experiments and online A/B
tests demonstrate the superiority of CQE in watch-time prediction and user
engagement modeling. Specifically, deploying CQE online on a large-scale
platform with hundreds of millions of daily active users has led to substantial
gains in key evaluation metrics, including active days, engagement time, and
video views. These results highlight the practical impact of our proposed
approach in enhancing the user experience and overall performance of the short
video recommendation system. The code will be released
https://github.com/justopit/CQE.
| [
{
"version": "v1",
"created": "Wed, 17 Jul 2024 00:25:35 GMT"
},
{
"version": "v2",
"created": "Sat, 27 Jul 2024 07:58:27 GMT"
},
{
"version": "v3",
"created": "Wed, 31 Jul 2024 12:49:17 GMT"
},
{
"version": "v4",
"created": "Sun, 13 Apr 2025 12:45:32 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Lin",
"Chengzhi",
""
],
[
"Liu",
"Shuchang",
""
],
[
"Wang",
"Chuyuan",
""
],
[
"Liu",
"Yongqi",
""
]
] |
2407.12769 | Daniel Durnford | M. M. Arora, L. Balogh, C. Beaufort, A. Brossard, M. Chapellier, J.
Clarke, E. C. Corcoran, J.-M. Coquillat, A. Dastgheibi-Fard, Y. Deng, D.
Durnford, C. Garrah, G. Gerbier, I. Giomataris, G. Giroux, P. Gorel, M. Gros,
P. Gros, O. Guillaudin, E. W. Hoppe, I. Katsioulas, F. Kelly, P. Knights, P.
Lautridou, A. Makowski, I. Manthos, R. D. Martin, J. Matthews, H. M.
McCallum, H. Meadows, L. Millins, J.-F. Muraz, T. Neep, K. Nikolopoulos, N.
Panchal, M.-C. Piro, N. Rowe, D. Santos, G. Savvidis, I. Savvidis, D.
Spathara, F. Vazquez de Sola Fernandez, R. Ward | Search for Light Dark Matter with NEWS-G at the Laboratoire Souterrain
de Modane Using a Methane Target | 9 pages, 5 figures | Phys. Rev. Lett. 134, 141002 (2025) | 10.1103/PhysRevLett.134.141002 | null | hep-ex astro-ph.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The NEWS-G direct detection experiment uses spherical proportional counters
to search for light dark matter candidates. New results from a 10 day physics
run with a $135\,\mathrm{cm}$ in diameter spherical proportional counter at the
Laboratoire Souterrain de Modane are reported. The target consists of
$114\,\mathrm{g}$ of methane, providing sensitivity to dark matter
spin-dependent coupling to protons. New constraints are presented in the mass
range $0.17$ to $1.2\,\mathrm{GeV/c^2}$, with a 90% confidence level
cross-section upper limit of $30.9\,\mathrm{pb}$ for a mass of
$0.76\,\mathrm{GeV/c^2}$.
| [
{
"version": "v1",
"created": "Wed, 17 Jul 2024 17:50:02 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Apr 2025 18:28:29 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Arora",
"M. M.",
""
],
[
"Balogh",
"L.",
""
],
[
"Beaufort",
"C.",
""
],
[
"Brossard",
"A.",
""
],
[
"Chapellier",
"M.",
""
],
[
"Clarke",
"J.",
""
],
[
"Corcoran",
"E. C.",
""
],
[
"Coquillat",
"J. -M.",
""
],
[
"Dastgheibi-Fard",
"A.",
""
],
[
"Deng",
"Y.",
""
],
[
"Durnford",
"D.",
""
],
[
"Garrah",
"C.",
""
],
[
"Gerbier",
"G.",
""
],
[
"Giomataris",
"I.",
""
],
[
"Giroux",
"G.",
""
],
[
"Gorel",
"P.",
""
],
[
"Gros",
"M.",
""
],
[
"Gros",
"P.",
""
],
[
"Guillaudin",
"O.",
""
],
[
"Hoppe",
"E. W.",
""
],
[
"Katsioulas",
"I.",
""
],
[
"Kelly",
"F.",
""
],
[
"Knights",
"P.",
""
],
[
"Lautridou",
"P.",
""
],
[
"Makowski",
"A.",
""
],
[
"Manthos",
"I.",
""
],
[
"Martin",
"R. D.",
""
],
[
"Matthews",
"J.",
""
],
[
"McCallum",
"H. M.",
""
],
[
"Meadows",
"H.",
""
],
[
"Millins",
"L.",
""
],
[
"Muraz",
"J. -F.",
""
],
[
"Neep",
"T.",
""
],
[
"Nikolopoulos",
"K.",
""
],
[
"Panchal",
"N.",
""
],
[
"Piro",
"M. -C.",
""
],
[
"Rowe",
"N.",
""
],
[
"Santos",
"D.",
""
],
[
"Savvidis",
"G.",
""
],
[
"Savvidis",
"I.",
""
],
[
"Spathara",
"D.",
""
],
[
"Fernandez",
"F. Vazquez de Sola",
""
],
[
"Ward",
"R.",
""
]
] |
2407.12778 | Girish Vishwa | Jos\'e Figueroa-O'Farrill and Girish S Vishwa | The BRST quantisation of chiral BMS-like field theories | v2: 28 pages. Fixed typos and added some references | J. Math. Phys. 1 April 2025; 66 (4): 042303 | 10.1063/5.0237868 | EMPG-24-3 | hep-th math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | The BMS$_3$ Lie algebra belongs to a one-parameter family of Lie algebras
obtained by centrally extending abelian extensions of the Witt algebra by a
tensor density representation. In this paper we call such Lie algebras
$\hat{\mathfrak{g}}_\lambda$, with BMS$_3$ corresponding to the universal
central extension of $\lambda = -1$. We construct the BRST complex for
$\hat{\mathfrak{g}}_\lambda$ in two different ways: one in the language of
semi-infinite cohomology and the other using the formalism of vertex operator
algebras. We pay particular attention to the case of BMS$_3$ and discuss some
natural field-theoretical realisations. We prove two theorems about the BRST
cohomology of $\hat{\mathfrak{g}}_\lambda$. The first is the construction of a
quasi-isomorphic embedding of the chiral sector of any Virasoro string as a
$\hat{\mathfrak{g}}_\lambda$ string. The second is the isomorphism (as
Batalin-Vilkovisky algebras) of any $\hat{\mathfrak{g}}_\lambda$ BRST
cohomology and the chiral ring of a topologically twisted $N{=}2$
superconformal field theory.
| [
{
"version": "v1",
"created": "Wed, 17 Jul 2024 17:56:42 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 15:35:37 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Figueroa-O'Farrill",
"José",
""
],
[
"Vishwa",
"Girish S",
""
]
] |
2407.13229 | Jindou Jia | Jindou Jia, Meng Wang, Zihan Yang, Bin Yang, Yuhang Liu, Kexin Guo,
Xiang Yu | Learning-based Observer for Coupled Disturbance | 17 pages, 9 figures | null | null | null | cs.RO cs.SY eess.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Achieving high-precision control for robotic systems is hindered by the
low-fidelity dynamical model and external disturbances. Especially, the
intricate coupling between internal uncertainties and external disturbances
further exacerbates this challenge. This study introduces an effective and
convergent algorithm enabling accurate estimation of the coupled disturbance
via combining control and learning philosophies. Concretely, by resorting to
Chebyshev series expansion, the coupled disturbance is firstly decomposed into
an unknown parameter matrix and two known structures dependent on system state
and external disturbance respectively. A regularized least squares algorithm is
subsequently formalized to learn the parameter matrix using historical
time-series data. Finally, a polynomial disturbance observer is specifically
devised to achieve a high-precision estimation of the coupled disturbance by
utilizing the learned portion. The proposed algorithm is evaluated through
extensive simulations and real flight tests. We believe this work can offer a
new pathway to integrate learning approaches into control frameworks for
addressing longstanding challenges in robotic applications.
| [
{
"version": "v1",
"created": "Thu, 18 Jul 2024 07:30:20 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 11:16:12 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Jia",
"Jindou",
""
],
[
"Wang",
"Meng",
""
],
[
"Yang",
"Zihan",
""
],
[
"Yang",
"Bin",
""
],
[
"Liu",
"Yuhang",
""
],
[
"Guo",
"Kexin",
""
],
[
"Yu",
"Xiang",
""
]
] |
2407.14181 | Bingqing Yao | Bingqing Yao, Chaokai Xu, Yaxin Tang, Yankun Du, Shengdong Tan, Sheng
Dai, Guangfu Luo, Qian He | Harnessing Zn-Volatility for Compositional Tuning in PtZn Nanoalloy
Catalysts | null | null | 10.1016/j.nantod.2025.102746 | null | physics.app-ph cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Bimetallic nanoalloys have gained extensive attention due to their tunable
properties and wide range of catalytic applications. However, achieving good
compositional control in nanoalloy catalysts remains a formidable challenge. In
this work, we demonstrate that heat treatment can be used to tune the
composition of Pt-Zn nanoalloy catalysts, leveraging the volatile nature of
zinc to enhance their performance in propane dehydrogenation. Through identical
location (scanning) transmission electron microscopy (IL-(S)TEM) using an
in-situ EM gas cell, as well as other complementary techniques, we observed
that the zinc content of the Pt-Zn nanoalloy particles decreased over time of
the heat treatment under hydrogen. The rate of change depends on the original
composition of the particles, as well as the heat treatment conditions such as
temperature and flow rate. Our experimental results and theoretical
calculations suggest that Zn in the intermetallic phase might be more stable,
providing an opportunity for precise tuning the nanoparticle compositions. This
approach presents a viable strategy for developing better Pt-Zn catalysts for
propane dehydrogenation.
| [
{
"version": "v1",
"created": "Fri, 19 Jul 2024 10:20:00 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Yao",
"Bingqing",
""
],
[
"Xu",
"Chaokai",
""
],
[
"Tang",
"Yaxin",
""
],
[
"Du",
"Yankun",
""
],
[
"Tan",
"Shengdong",
""
],
[
"Dai",
"Sheng",
""
],
[
"Luo",
"Guangfu",
""
],
[
"He",
"Qian",
""
]
] |
2407.14919 | Guodong Jiang | Guodong Jiang, P\"aivi T\"orm\"a, and Yafis Barlas | Superfluid weight cross-over and critical temperature enhancement in
singular flat bands | null | null | 10.1073/pnas.2416726122 | null | cond-mat.supr-con | http://creativecommons.org/licenses/by/4.0/ | Non-analytic Bloch eigenstates at isolated band degeneracy points exhibit
singular behavior in the quantum metric. Here, a description of superfluid
weight for zero-energy flat bands in proximity to other high-energy bands is
presented, where they together form a singular band gap system. When the
singular band gap closes, the geometric and conventional contributions to the
superfluid weight as a function of the superconducting gap exhibit different
crossover behaviors. The scaling behavior of superfluid weight with the band
gap is studied in detail, and the effect on the Berezinskii-KosterlitzThouless
(BKT) transition temperature is explored. It is discovered that tuning the
singular band gap provides a unique mechanism for enhancing the supercurrent
and critical temperature of two-dimensional (2D) superconductors.
| [
{
"version": "v1",
"created": "Sat, 20 Jul 2024 16:11:45 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 00:41:47 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Jiang",
"Guodong",
""
],
[
"Törmä",
"Päivi",
""
],
[
"Barlas",
"Yafis",
""
]
] |
2407.14927 | Aleksei Tsarapkin | Aleksei Tsarapkin, Luka Zurak, Krzysztof Ma\'ckosz, Lorenz L\"offler,
Victor Deinhart, Ivo Utke, Thorsten Feichtner, Katja H\"oflich | Double helical plasmonic antennas | 17 pages with Supporting Information, 7 figures | A. Tsarapkin, L. Zurak, K. Ma\'ckosz, L. L\"offler, V. Deinhart,
I. Utke, T. Feichtner, K. H\"oflich, Double Helical Plasmonic Antennas. Adv.
Funct. Mater. 2025, 2507471 | 10.1002/adfm.202507471 | null | physics.optics physics.app-ph | http://creativecommons.org/licenses/by/4.0/ | Plasmonic double helical antennas are a means to funnel circularly polarized
light down to the nanoscale. Here, an existing design tool for single helices
is extended to the case of double helices and used to design antennas that
combine large chiroptical interaction strength with highly directional light
emission. Full-field numerical modeling underpins the design and provides
additional insight into surface charge distributions and resonance widths. The
helical antennas are fabricated by direct writing with a focused electron beam,
a technique that is unrivaled in terms of spatial resolution and 3D shape
fidelity. After the printing process, the structures are purified using ozone
plasma at room temperature, resulting in the smallest continuous double helix
antennas ever realized in gold. Fabricated antennas are studied regarding their
polarization-dependent transmission behavior, which shows a large and broadband
dissymmetry factor in the visible range. Since the polarization of light is an
important tool for implementing logic functionality in photonic and quantum
photonic devices, these helices are potential building blocks for future
nanophotonic circuits, but also for chiral metamaterials or phase plates.
| [
{
"version": "v1",
"created": "Sat, 20 Jul 2024 16:27:18 GMT"
},
{
"version": "v2",
"created": "Wed, 14 Aug 2024 14:20:19 GMT"
},
{
"version": "v3",
"created": "Mon, 14 Apr 2025 09:24:36 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Tsarapkin",
"Aleksei",
""
],
[
"Zurak",
"Luka",
""
],
[
"Maćkosz",
"Krzysztof",
""
],
[
"Löffler",
"Lorenz",
""
],
[
"Deinhart",
"Victor",
""
],
[
"Utke",
"Ivo",
""
],
[
"Feichtner",
"Thorsten",
""
],
[
"Höflich",
"Katja",
""
]
] |
2407.15170 | Zhu Huang | Zhu Huang, Gang Pan, Chao Kang, YaoZhi Lv | Semi-Supervised Pipe Video Temporal Defect Interval Localization | 13 pages, 3 figures | null | 10.1111/mice.13403 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In sewer pipe Closed-Circuit Television (CCTV) inspection, accurate temporal
defect localization is essential for effective defect classification,
detection, segmentation and quantification. Industry standards typically do not
require time-interval annotations, even though they are more informative than
time-point annotations for defect localization, resulting in additional
annotation costs when fully supervised methods are used. Additionally,
differences in scene types and camera motion patterns between pipe inspections
and Temporal Action Localization (TAL) hinder the effective transfer of
point-supervised TAL methods. Therefore, this study introduces a
Semi-supervised multi-Prototype-based method incorporating visual Odometry for
enhanced attention guidance (PipeSPO). PipeSPO fully leverages unlabeled data
through unsupervised pretext tasks and utilizes time-point annotated data with
a weakly supervised multi-prototype-based method, relying on visual odometry
features to capture camera pose information. Experiments on real-world datasets
demonstrate that PipeSPO achieves 41.89% average precision across Intersection
over Union (IoU) thresholds of 0.1-0.7, improving by 8.14% over current
state-of-the-art methods.
| [
{
"version": "v1",
"created": "Sun, 21 Jul 2024 14:04:21 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Huang",
"Zhu",
""
],
[
"Pan",
"Gang",
""
],
[
"Kang",
"Chao",
""
],
[
"Lv",
"YaoZhi",
""
]
] |
2407.15735 | Alexandre Matov | Alexandre Matov | Real-Time Image Analysis Software Suitable for Resource-Constrained
Computing | null | null | null | null | q-bio.QM | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Methods: We have developed a software suite (DataSet Tracker) for real-time
analysis designed to run on computers, smartphones, and smart glasses hardware
and suitable for resource-constrained, on-the-fly computing in microscopes
without internet connectivity; a demo is available for viewing at
datasetanalysis.com. Our objective is to present the community with an
integrated, easy to use by all, tool for resolving the complex dynamics of the
cytoskeletal meshworks, intracytoplasmic membranous networks, and vesicle
trafficking. Our software is optimized for resource-constrained computing and
can be installed even on microscopes without internet connectivity.
Results: Our computational platform can provide high-content analyses and
functional secondary screening of novel compounds that are in the process of
approval, or at a pre-clinical stage of development, and putative combination
therapies based on FDA-approved drugs. Importantly, dissecting the mechanisms
of drug action with quantitative detail will allow the design of drugs that
impede relapse and optimal dose regimens with minimal harmful side effects by
carefully exploiting disease-specific aberrations.
Conclusions: DataSet Tracker, the real-time optical flow feature tracking
software presented in this contribution, can serve as the base module of an
integrated platform of existing and future algorithms for real-time cellular
analysis. The computational assay we propose could successfully be applied to
evaluate treatment strategies for any human organ. It is our goal to have this
integrated tool approved for use in the clinical practice.
| [
{
"version": "v1",
"created": "Mon, 22 Jul 2024 15:39:58 GMT"
},
{
"version": "v2",
"created": "Tue, 23 Jul 2024 05:53:03 GMT"
},
{
"version": "v3",
"created": "Fri, 16 Aug 2024 16:57:31 GMT"
},
{
"version": "v4",
"created": "Tue, 19 Nov 2024 21:38:28 GMT"
},
{
"version": "v5",
"created": "Tue, 10 Dec 2024 21:08:07 GMT"
},
{
"version": "v6",
"created": "Thu, 16 Jan 2025 04:18:47 GMT"
},
{
"version": "v7",
"created": "Sat, 1 Mar 2025 20:38:56 GMT"
},
{
"version": "v8",
"created": "Sun, 30 Mar 2025 01:20:43 GMT"
},
{
"version": "v9",
"created": "Sat, 12 Apr 2025 03:10:06 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Matov",
"Alexandre",
""
]
] |
2407.16682 | Pengfei Chen | Pengfei Chen, Lingxi Xie, Xinyue Huo, Xuehui Yu, Xiaopeng Zhang,
Yingfei Sun, Zhenjun Han, Qi Tian | SAM-CP: Marrying SAM with Composable Prompts for Versatile Segmentation | Accepted by ICLR 2025; codes:https://github.com/ucas-vg/SAM-CP | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Segment Anything model (SAM) has shown a generalized ability to group
image pixels into patches, but applying it to semantic-aware segmentation still
faces major challenges. This paper presents SAM-CP, a simple approach that
establishes two types of composable prompts beyond SAM and composes them for
versatile segmentation. Specifically, given a set of classes (in texts) and a
set of SAM patches, the Type-I prompt judges whether a SAM patch aligns with a
text label, and the Type-II prompt judges whether two SAM patches with the same
text label also belong to the same instance. To decrease the complexity in
dealing with a large number of semantic classes and patches, we establish a
unified framework that calculates the affinity between (semantic and instance)
queries and SAM patches and merges patches with high affinity to the query.
Experiments show that SAM-CP achieves semantic, instance, and panoptic
segmentation in both open and closed domains. In particular, it achieves
state-of-the-art performance in open-vocabulary segmentation. Our research
offers a novel and generalized methodology for equipping vision foundation
models like SAM with multi-grained semantic perception abilities.
| [
{
"version": "v1",
"created": "Tue, 23 Jul 2024 17:47:25 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 05:49:03 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Chen",
"Pengfei",
""
],
[
"Xie",
"Lingxi",
""
],
[
"Huo",
"Xinyue",
""
],
[
"Yu",
"Xuehui",
""
],
[
"Zhang",
"Xiaopeng",
""
],
[
"Sun",
"Yingfei",
""
],
[
"Han",
"Zhenjun",
""
],
[
"Tian",
"Qi",
""
]
] |
2407.17107 | Weihua Yang | W. Yang and C. Li | The nuclear matter effect on parton distribution functions | null | null | null | null | nucl-th hep-ph | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Parton distribution functions (PDFs) are important quantities in describing
nucleon structures. They are universal and process-independent. As a matter of
fact, nucleon PDFs are inevitably affected by nuclear matter during nuclear
scattering process. In order to study the nuclear PDFs (nPDFs), in this paper,
we introduce the nucleon pair PDFs (dPDFs) to describe parton distributions in
the nucleon pair which is confined to a nucleus. We first of all construct the
nuclear state in terms of the nucleonic states and calculate the operator
definition of nPDFs. Neglecting the higher order corrections or multi-nucleon
correlations, we find that nPDFs can be written as a sum of two terms which
respectively correspond to PDFs and dPDFs. Nucleon pair PDFs which stem from
nucleon-nucleon correlation are proportional to common nucleon PDFs but
suppressed by a factor. Compared with the experimental data, we find that dPDFs
can explain the behaviour of the EMC effect. We further calculate the
Paschos-Wolfenstein ratio to study the nuclear matter effect on the extraction
of weak mixing angle or $\sin^2\theta_W$ by using dPDFs.
| [
{
"version": "v1",
"created": "Wed, 24 Jul 2024 09:07:10 GMT"
},
{
"version": "v2",
"created": "Sat, 27 Jul 2024 04:58:13 GMT"
},
{
"version": "v3",
"created": "Mon, 14 Apr 2025 08:11:44 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Yang",
"W.",
""
],
[
"Li",
"C.",
""
]
] |
2407.18055 | Uesli Alushi | Uesli Alushi, Alessandro Coppo, Valentina Brosco, Roberto Di Candia,
Simone Felicetti | Collective quantum enhancement in critical quantum sensing | 17 pages, 5 figures. Final version | Commun. Phys. 8, 74 (2025) | 10.1038/s42005-025-01975-9 | null | quant-ph cond-mat.mes-hall | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Critical systems represent a valuable resource in quantum sensing and
metrology. Critical quantum sensing (CQS) protocols can be realized using
finite-component phase transitions, where criticality arises from the rescaling
of system parameters rather than the thermodynamic limit. Here, we show that a
collective quantum advantage can be achieved in a multipartite CQS protocol
using a chain of parametrically coupled critical resonators in the
weak-nonlinearity limit. We derive analytical solutions for the low-energy
spectrum of this unconventional quantum many-body system, which is composed of
locally critical elements. We then assess the scaling of the quantum Fisher
information with respect to fundamental resources. We demonstrate that the
coupled chain outperforms an equivalent ensemble of independent critical
sensors, achieving quadratic scaling in the number of resonators. Finally, we
show that even with finite Kerr nonlinearity or Markovian dissipation, the
critical chain retains its advantage, making it relevant for implementing
quantum sensors with current microwave superconducting technologies.
| [
{
"version": "v1",
"created": "Thu, 25 Jul 2024 14:08:39 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 09:30:51 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Alushi",
"Uesli",
""
],
[
"Coppo",
"Alessandro",
""
],
[
"Brosco",
"Valentina",
""
],
[
"Di Candia",
"Roberto",
""
],
[
"Felicetti",
"Simone",
""
]
] |
2407.18321 | Wies{\l}aw Sobk\'ow | A. B{\l}aut (1), W. Sobk\'ow (1) ((1) Institute of Theoretical
Physics, University of Wroc{\l}aw, Poland) | Recoil electron polarization-dependent T-odd correlations in neutrino
elastic scattering on polarized electron target | 18 pages, 1 figure, changed title, modifications in abstract,
introduction and conclusions | null | 10.1140/epjc/s10052-025-14084-2 | null | hep-ph | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Possible symmetry breaking tests with respect to the time inversion in the
elastic scattering of neutrinos on polarized electrons are considered, assuming
that the incoming neutrino beam is either longitudinally or transversely
polarized, and both momentum and polarization of recoil electrons are observed.
In the process, in addition to the standard interaction, the exotic scalar,
pseudoscalar and tensor interactions can participate. Due to the nonstandard
interactions, different types of triple T-odd products in the cross section may
appear. We consider several experiments in which mixed products built of the
recoil electron polarization and two other vector quantities (incoming neutrino
momentum, its polarization, polarization of the electron target and outgoing
electron momentum) can be identified.
Observations of the T-odd correlations may indicate noninvariance under time
reversal as well as the presence of exotic interactions. A complete analysis of
the problem requires a precise determination of the possible contributions from
the interactions mimicking the genuine violation of time reversal symmetry,
e.g. final state interactions.
| [
{
"version": "v1",
"created": "Thu, 25 Jul 2024 18:22:49 GMT"
},
{
"version": "v2",
"created": "Sat, 14 Dec 2024 09:01:22 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Błaut",
"A.",
""
],
[
"Sobków",
"W.",
""
]
] |
2407.18449 | Jiabo Ma | Jiabo Ma, Zhengrui Guo, Fengtao Zhou, Yihui Wang, Yingxue Xu, Jinbang
Li, Fang Yan, Yu Cai, Zhengjie Zhu, Cheng Jin, Yi Lin, Xinrui Jiang,
Chenglong Zhao, Danyi Li, Anjia Han, Zhenhui Li, Ronald Cheong Kin Chan,
Jiguang Wang, Peng Fei, Kwang-Ting Cheng, Shaoting Zhang, Li Liang, Hao Chen | Towards A Generalizable Pathology Foundation Model via Unified Knowledge
Distillation | update | null | null | I.2.10 | eess.IV cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Foundation models pretrained on large-scale datasets are revolutionizing the
field of computational pathology (CPath). The generalization ability of
foundation models is crucial for the success in various downstream clinical
tasks. However, current foundation models have only been evaluated on a limited
type and number of tasks, leaving their generalization ability and overall
performance unclear. To address this gap, we established a most comprehensive
benchmark to evaluate the performance of off-the-shelf foundation models across
six distinct clinical task types, encompassing a total of 72 specific tasks,
including slide-level classification, survival prediction, ROI-tissue
classification, ROI retrieval, visual question answering, and report
generation. Our findings reveal that existing foundation models excel at
certain task types but struggle to effectively handle the full breadth of
clinical tasks. To improve the generalization of pathology foundation models,
we propose a unified knowledge distillation framework consisting of both expert
and self-knowledge distillation, where the former allows the model to learn
from the knowledge of multiple expert models, while the latter leverages
self-distillation to enable image representation learning via local-global
alignment. Based on this framework, we curated a dataset of 96,000 whole slide
images (WSIs) and developed a Generalizable Pathology Foundation Model (GPFM).
This advanced model was trained on a substantial dataset comprising 190 million
images extracted from approximately 72,000 publicly available slides,
encompassing 34 major tissue types. Evaluated on the established benchmark,
GPFM achieves an impressive average rank of 1.6, with 42 tasks ranked 1st,
while the second-best model, UNI, attains an average rank of 3.7, with only 6
tasks ranked 1st.
| [
{
"version": "v1",
"created": "Fri, 26 Jul 2024 01:12:54 GMT"
},
{
"version": "v2",
"created": "Sat, 3 Aug 2024 13:36:24 GMT"
},
{
"version": "v3",
"created": "Mon, 14 Apr 2025 09:55:21 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Ma",
"Jiabo",
""
],
[
"Guo",
"Zhengrui",
""
],
[
"Zhou",
"Fengtao",
""
],
[
"Wang",
"Yihui",
""
],
[
"Xu",
"Yingxue",
""
],
[
"Li",
"Jinbang",
""
],
[
"Yan",
"Fang",
""
],
[
"Cai",
"Yu",
""
],
[
"Zhu",
"Zhengjie",
""
],
[
"Jin",
"Cheng",
""
],
[
"Lin",
"Yi",
""
],
[
"Jiang",
"Xinrui",
""
],
[
"Zhao",
"Chenglong",
""
],
[
"Li",
"Danyi",
""
],
[
"Han",
"Anjia",
""
],
[
"Li",
"Zhenhui",
""
],
[
"Chan",
"Ronald Cheong Kin",
""
],
[
"Wang",
"Jiguang",
""
],
[
"Fei",
"Peng",
""
],
[
"Cheng",
"Kwang-Ting",
""
],
[
"Zhang",
"Shaoting",
""
],
[
"Liang",
"Li",
""
],
[
"Chen",
"Hao",
""
]
] |
2407.19566 | Sanaz Mahmoodi Takaghaj | Sanaz Mahmoodi Takaghaj, Jack Sampson | To Spike or Not to Spike, that is the Question | Accepted to Artificial Intelligence Circuits And Systems(AICAS), 2015 | null | null | null | cs.ET cs.NE | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Neuromorphic computing has recently gained momentum with the emergence of
various neuromorphic processors. As the field advances, there is an increasing
focus on developing training methods that can effectively leverage the unique
properties of spiking neural networks (SNNs). SNNs emulate the temporal
dynamics of biological neurons, making them particularly well-suited for
real-time, event-driven processing. To fully harness the potential of SNNs
across different neuromorphic platforms, effective training methodologies are
essential. In SNNs, learning rules are based on neurons' spiking behavior, that
is, if and when spikes are generated due to a neuron's membrane potential
exceeding that neuron's spiking threshold, and this spike timing encodes vital
information. However, the threshold is generally treated as a hyperparameter,
and incorrect selection can lead to neurons that do not spike for large
portions of the training process, hindering the effective rate of learning.
This work focuses on the significance of learning neuron thresholds alongside
weights in SNNs. Our results suggest that promoting threshold from a
hyperparameter to a trainable parameter effectively addresses the issue of dead
neurons during training. This leads to a more robust training algorithm,
resulting in improved convergence, increased test accuracy, and a substantial
reduction in the number of training epochs required to achieve viable accuracy
on spatiotemporal datasets such as NMNIST, DVS128, and Spiking Heidelberg
Digits (SHD), with up to 30% training speed-up and up to 2% higher accuracy on
these datasets.
| [
{
"version": "v1",
"created": "Sun, 28 Jul 2024 19:23:09 GMT"
},
{
"version": "v2",
"created": "Thu, 31 Oct 2024 19:45:37 GMT"
},
{
"version": "v3",
"created": "Sun, 13 Apr 2025 01:39:25 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Takaghaj",
"Sanaz Mahmoodi",
""
],
[
"Sampson",
"Jack",
""
]
] |
2407.20021 | Kanghyun Choi | Kanghyun Choi, Hye Yoon Lee, Dain Kwon, SunJong Park, Kyuyeun Kim,
Noseong Park, Jonghyun Choi, Jinho Lee | MimiQ: Low-Bit Data-Free Quantization of Vision Transformers with
Encouraging Inter-Head Attention Similarity | Published to AAAI 2025 | null | 10.1609/aaai.v39i15.33761 | null | cs.LG cs.AI cs.CV | http://creativecommons.org/licenses/by/4.0/ | Data-free quantization (DFQ) is a technique that creates a lightweight
network from its full-precision counterpart without the original training data,
often through a synthetic dataset. Although several DFQ methods have been
proposed for vision transformer (ViT) architectures, they fail to achieve
efficacy in low-bit settings. Examining the existing methods, we observe that
their synthetic data produce misaligned attention maps, while those of the real
samples are highly aligned. From this observation, we find that aligning
attention maps of synthetic data helps improve the overall performance of
quantized ViTs. Motivated by this finding, we devise MimiQ, a novel DFQ method
designed for ViTs that enhances inter-head attention similarity. First, we
generate synthetic data by aligning head-wise attention outputs from each
spatial query patch. Then, we align the attention maps of the quantized network
to those of the full-precision teacher by applying head-wise structural
attention distillation. The experimental results show that the proposed method
significantly outperforms baselines, setting a new state-of-the-art for
ViT-DFQ. This paper is an extended version of our work published in the
proceedings of AAAI 2025, including additional supplementary material.
| [
{
"version": "v1",
"created": "Mon, 29 Jul 2024 13:57:40 GMT"
},
{
"version": "v2",
"created": "Tue, 30 Jul 2024 02:03:06 GMT"
},
{
"version": "v3",
"created": "Thu, 1 Aug 2024 16:13:45 GMT"
},
{
"version": "v4",
"created": "Mon, 14 Apr 2025 07:42:52 GMT"
}
] | 2025-04-15T00:00:00 | [
[
"Choi",
"Kanghyun",
""
],
[
"Lee",
"Hye Yoon",
""
],
[
"Kwon",
"Dain",
""
],
[
"Park",
"SunJong",
""
],
[
"Kim",
"Kyuyeun",
""
],
[
"Park",
"Noseong",
""
],
[
"Choi",
"Jonghyun",
""
],
[
"Lee",
"Jinho",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.