id
string | submitter
string | authors
string | title
string | comments
string | journal-ref
string | doi
string | report-no
string | categories
string | license
string | abstract
string | versions
list | update_date
timestamp[s] | authors_parsed
sequence |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2405.02199 | Dominic Bernardi | Dominic Bernardi, Yajie Yuan, Alexander Y. Chen (Washington University in St. Louis) | Alfv\'en Wave Conversion to Low Frequency Fast Magnetosonic Waves in Magnetar Magnetospheres | 12 pages, 13 figures | null | null | null | astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Rapid shear motion of magnetar crust can launch Alfv\'{e}n waves into the magnetosphere. The dissipation of the Alfv\'{e}n waves has been theorized to power the X-ray bursts characteristic of magnetars. However, the process by which Alfv\'{e}n waves convert their energy to X-rays is unclear. Recent work has suggested that energetic fast magnetosonic (fast) waves can be produced as a byproduct of Alfv\'{e}n waves propagating on curved magnetic field lines; their subsequent dissipation may power X-ray bursts. In this work, we investigate the production of fast waves by performing axisymmetric force-free simulations of Alfv\'{e}n waves propagating in a dipolar magnetosphere. For Alfv\'{e}n wave trains that do not completely fill the flux tube confining them, we find a fast wave dominated by a low frequency component with a wavelength defined by the bouncing time of the Alfv\'{e}n waves. In contrast, when the wave train is long enough to completely fill the flux tube, and the Alfv\'{e}n waves overlap significantly, the energy is quickly converted into a fast wave with a higher frequency that corresponds to twice the Alfv\'{e}n wave frequency. We investigate how the energy, duration, and wavelength of the initial Alfv\'{e}n wave train affect the conversion efficiency to fast waves. For modestly energetic star quakes, we see that the fast waves that are produced will become non-linear well within the magnetosphere, and we comment on the X-ray emission that one may expect from such events. | [
{
"version": "v1",
"created": "Fri, 03 May 2024 15:57:16 GMT"
},
{
"version": "v2",
"created": "Fri, 06 Jun 2025 22:35:28 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Bernardi",
"Dominic",
"",
"Washington University in St. Louis"
],
[
"Yuan",
"Yajie",
"",
"Washington University in St. Louis"
],
[
"Chen",
"Alexander Y.",
"",
"Washington University in St. Louis"
]
] |
2405.02430 | Yisen Wang | Shaoshi Chen, Christoph Koutschan, Yisen Wang | How to generate all possible rational Wilf-Zeilberger forms? | null | null | null | null | cs.SC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Wilf-Zeilberger pairs are fundamental in the algorithmic theory of Wilf and Zeilberger for computer-generated proofs of combinatorial identities. Wilf-Zeilberger forms are their high-dimensional generalizations, which can be used for proving and discovering convergence acceleration formulas. This paper presents a structural description of all possible rational such forms, which can be viewed as an additive analog of the classical Ore-Sato theorem. Based on this analog, we show a structural decomposition of so-called multivariate hyperarithmetic terms, which extend multivariate hypergeometric terms to the additive setting. | [
{
"version": "v1",
"created": "Fri, 03 May 2024 18:52:23 GMT"
},
{
"version": "v2",
"created": "Sun, 08 Jun 2025 03:31:43 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Chen",
"Shaoshi",
""
],
[
"Koutschan",
"Christoph",
""
],
[
"Wang",
"Yisen",
""
]
] |
2405.03282 | Lenny Neyt | Lenny Neyt, Joachim Toft, Jasson Vindas | Hermite expansions for spaces of functions with nearly optimal time-frequency decay | 14 pages | J. Funct. Anal. 288 (2025), Article 110706 | 10.1016/j.jfa.2024.110706 | null | math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We establish Hermite expansion characterizations for several subspaces of the Fr\'{e}chet space of functions on the real line satisfying \begin{equation*} |f(x)| \lesssim e^{-(\frac{1}{2} - \lambda ) x^{2}} , \qquad | \widehat{f}(\xi )| \lesssim e^{-(\frac{1}{2} - \lambda ) \xi ^{2}} , \qquad \forall \lambda > 0 . \end{equation*} In particular, we extend and improve Fourier characterizations of the so-called proper Pilipovi\'{c} spaces obtained in [J. Funct. Anal. 284 (2023), 109724]. The main ingredients in our proofs are the Bargmann transform and some achieved optimal forms of the Phragm\'{e}n-Lindel\"{o}f principle. | [
{
"version": "v1",
"created": "Mon, 06 May 2024 08:57:58 GMT"
},
{
"version": "v2",
"created": "Fri, 25 Oct 2024 09:05:37 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Neyt",
"Lenny",
""
],
[
"Toft",
"Joachim",
""
],
[
"Vindas",
"Jasson",
""
]
] |
2405.03437 | Andreas Wurzinger | Andreas Wurzinger and Stefan Schoder | pyCFS-data: Data Processing Framework in Python for openCFS | null | null | null | null | cs.CE cs.NA math.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many numerical simulation tools have been developed and are on the market, but there is still a strong need for appropriate tools capable of simulating multi-field problems, especially in aeroacoustics. Therefore, openCFS provides an open-source framework for implementing partial differential equations using the finite element method. Since 2000, the software has been developed continuously. The result is openCFS (before 2020, known as CFS++ Coupled Field Simulations written in C++). In this paper, we present pyCFS-data, a data processing framework written in Python to provide a flexible and easy-to-use toolbox to access and manipulate, pre- and postprocess data generated by or for usage with openCFS. | [
{
"version": "v1",
"created": "Mon, 06 May 2024 13:02:47 GMT"
},
{
"version": "v2",
"created": "Sun, 08 Jun 2025 16:48:13 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Wurzinger",
"Andreas",
""
],
[
"Schoder",
"Stefan",
""
]
] |
2405.03750 | Fan Yang | Fan Yang and Emil J. Bergholtz | Anatomy of Higher-Order Non-Hermitian Skin and Boundary Modes | 28 pages, 14 figures | Phys. Rev. Research 7, 023233 (2025) | 10.1103/PhysRevResearch.7.023233 | null | cond-mat.mes-hall physics.optics quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The anomalous bulk-boundary correspondence in non-Hermitian systems featuring an intricate interplay between skin and boundary modes has attracted enormous theoretical and experimental attention. Still, in dimensions higher than one, this interplay remains much less understood. Here we provide insights from exact analytical solutions of a large class of models in any dimension, $d$, with open boundaries in $d_c \le d$ directions and by tracking their topological origin. Specifically, we show that amoeba theory accounting for the separation gaps of the bulk modes augmented with higher-dimensional generalizations of the biorthogonal polarization and the generalized Brillouin zone approaches accounting for the surface gaps of boundary modes provide a comprehensive understanding of these systems. | [
{
"version": "v1",
"created": "Mon, 06 May 2024 18:00:01 GMT"
},
{
"version": "v2",
"created": "Tue, 18 Feb 2025 14:45:13 GMT"
},
{
"version": "v3",
"created": "Sun, 08 Jun 2025 15:46:20 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Yang",
"Fan",
""
],
[
"Bergholtz",
"Emil J.",
""
]
] |
2405.04884 | Zizhu Wang | Kaiyan Yang, Yanzheng Zhu, Xiao Zeng, Zuoheng Zou, Man-Hong Yung and Zizhu Wang | Cost of Locally Approximating High-Dimensional Ground States of Contextual Quantum Models | 11 pages, 6 figures, including one figure spanning two pages | Communications Physics 8, 204 (2025) | 10.1038/s42005-025-02127-9 | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Contextuality, one of the strongest forms of quantum correlations, delineates the quantum world and the classical one. It has been shown recently that some quantum models, in the form of infinite one-dimensional translation-invariant Hamiltonians with nearest- and next-to-nearest-neighbor interactions, have the lowest ground state energy density allowed in quantum physics. However, these models all have local Hilbert space dimension larger than two, making the study of their ground state behavior difficult on current qubit-based variational quantum simulation platforms. In this work, we focus on the cost of simulating the local approximations of ground states of these models using qubit-based parameterized quantum circuits. The local approximations, which are 3-site reduced density matrices with local Hilbert space dimension three, are purified then encoded into permutation-symmetric qubits. We develop a universal set of permutation-symmetry preserving qubit-based gates, using them as an ansatz to simulate parameterized quantum circuits designed for qutrits. These techniques allow us to assess the accuracy of simulating the purified local ground states with respect to a fixed amount of classical and quantum resources. We found that given the same quantum circuit and the number of iterations, more contextual ground states with lower energy density are easier to simulate. | [
{
"version": "v1",
"created": "Wed, 08 May 2024 08:35:31 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 08:32:17 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Yang",
"Kaiyan",
""
],
[
"Zhu",
"Yanzheng",
""
],
[
"Zeng",
"Xiao",
""
],
[
"Zou",
"Zuoheng",
""
],
[
"Yung",
"Man-Hong",
""
],
[
"Wang",
"Zizhu",
""
]
] |
2405.04989 | Swanhild Bernstein | Swanhild Bernstein and Nelson Faustino | Paley-Wiener Type Theorems associated to Dirac Operators of Riesz-Feller type | 33 pages, revised version, some minor changes | null | null | null | math.CV math.FA | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper explores Paley-Wiener type theorems within the framework of hypercomplex variables. The investigation focuses on a space-fractional version of the Dirac operator $\mathbf{D}_\theta^{\alpha}$ of order $\alpha$ and skewness $\theta$. The pseudo-differential reformulation of $\mathbf{D}_\theta^{\alpha}$ in terms of the Riesz derivative $(-\Delta)^{\frac{\alpha}{2}}$ and the so-called {\textit Riesz-Hilbert transform} $H$, allows for the description of generalized Hardy spaces on the upper and lower half-spaces of $\mathbf{R}^{n+1}$, $\mathbf{R}^{n+1}_+$ resp. $\mathbb{R}^{n+1}_-$, using L\'evy-Feller type semigroups generated by $-(-\Delta)^{\frac{\alpha}{2}}$, and the boundary values $\mathbf{f}_\pm=\frac{1}{2}\left(\mathbf{f}\pm H\mathbf{f}\right)$.
Subsequently, we employ a proof strategy rooted in {\textit real Paley-Wiener methods} to demonstrate that the growth behavior of the sequences of functions $\left(\left(\mathbf{D}_\theta^{\alpha}\right)^k\mathbf{f}_{\pm}\right)_{k\in \mathbb{N}_0}$ effectively captures the relationship between the support of the Fourier transform $\widehat{\mathbf{f}}$ of the $L^p-$function $\mathbf{f}$, in the case where $\mathrm{supp}\widehat{\mathbf{f}}\subseteq \overline{B(0,R)}$, and the solutions of Cauchy problems equipped with the space-time operator $\partial_{x_0} + \mathbf{D}_\theta^{\alpha}$, which are of exponential type $R^\alpha$.
Within the developed framework, introducing a hypercomplex analog for the Bernstein spaces $B_R^p$ arises naturally, allowing for the meaningful extension of the results by Kou and Qian as well as Franklin, Hogan, and Larkin. Specifically, leveraging the established Stein-Kolmogorov inequalities for hypercomplex variables enables us to accurately determine the maximum radius $R$ for which $\operatorname{supp}\widehat{\mathbf{f}} \subseteq \overline{B(0, R)}$ holds. | [
{
"version": "v1",
"created": "Wed, 08 May 2024 11:53:43 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Apr 2025 13:00:09 GMT"
},
{
"version": "v3",
"created": "Sat, 07 Jun 2025 17:25:12 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Bernstein",
"Swanhild",
""
],
[
"Faustino",
"Nelson",
""
]
] |
2405.05039 | Adrian Azzarelli | Adrian Azzarelli, Nantheera Anantrasirichai, David R Bull | Reviewing Intelligent Cinematography: AI research for camera-based video production | This paper has been accepted for publication with "Artificial Intelligence Review" Journal (https://link.springer.com/journal/10462) and we are in the procress of publishing it | null | 10.1007/s10462-024-11089-3 | null | cs.CV cs.MM | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper offers the first comprehensive review of artificial intelligence (AI) research in the context of real camera content acquisition for entertainment purposes and is aimed at both researchers and cinematographers. Addressing the lack of review papers in the field of intelligent cinematography} (IC) and the breadth of related computer vision research, we present a holistic view of the IC landscape while providing technical insight, important for experts across disciplines. We provide technical background on generative AI, object detection, automated camera calibration and 3-D content acquisition, with references to assist non-technical readers. The application sections categorize work in terms of four production types: General Production, Virtual Production, Live Production and Aerial Production. Within each application section, we (1) sub-classify work according to research topic and (2) describe the trends and challenges relevant to each type of production. In the final chapter, we address the greater scope of IC research and summarize the significant potential of this area to influence the creative industries sector. We suggest that work relating to virtual production has the greatest potential to impact other mediums of production, driven by the growing interest in LED volumes/stages for in-camera virtual effects (ICVFX) and automated 3-D capture for virtual modeling of real world scenes and actors. We also address ethical and legal concerns regarding the use of creative AI that impact on artists, actors, technologists and the general public. | [
{
"version": "v1",
"created": "Wed, 08 May 2024 13:13:02 GMT"
},
{
"version": "v2",
"created": "Mon, 23 Dec 2024 11:38:01 GMT"
},
{
"version": "v3",
"created": "Mon, 06 Jan 2025 14:41:22 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Azzarelli",
"Adrian",
""
],
[
"Anantrasirichai",
"Nantheera",
""
],
[
"Bull",
"David R",
""
]
] |
2405.05047 | Thomas Richter | Manuel Liebchen, Robert Jendersie, Utku Kaya, Christian Lessig, Thomas Richter | An adaptive finite element multigrid solver using GPU acceleration | null | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by-sa/4.0/ | Adaptive finite elements combined with geometric multigrid solvers are one of the most efficient numerical methods for problems such as the instationary Navier-Stokes equations. Yet despite their efficiency, computations remain expensive and the simulation of, for example, complex flow problems can take many hours or days. GPUs provide an interesting avenue to speed up the calculations due to their very large theoretical performance. However, the large degree of parallelism and non-standard API make the use of GPUs in scientific computing challenging. In this work, we develop a GPU acceleration for the adaptive finite element library Gascoigne and study its effectiveness for different systems of partial differential equations. Our goal is thereby to integrate the GPU acceleration into the existing code with minimal changes, even when this requires a penalty in the GPU acceleration. Through the systematic formulation of all computations as linear algebra operations, we can employ GPU-accelerated linear algebra libraries, which simplifies the implementation and ensures the maintainability of the code while achieving very efficient GPU utilizations. Our results for a transport-diffusion equation, linear elasticity, and the instationary Navier-Stokes equations show substantial speedups of up to 20X compared to multi-core CPU implementations. | [
{
"version": "v1",
"created": "Wed, 08 May 2024 13:35:01 GMT"
},
{
"version": "v2",
"created": "Fri, 06 Jun 2025 18:53:15 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Liebchen",
"Manuel",
""
],
[
"Jendersie",
"Robert",
""
],
[
"Kaya",
"Utku",
""
],
[
"Lessig",
"Christian",
""
],
[
"Richter",
"Thomas",
""
]
] |
2405.06314 | Daniel Fatu{\l}a | Daniel Fatu{\l}a | Applications of the Painlev\'e-Kuratowski convergence: Lipschitz functions with converging Clarke subdifferentials and convergence of sets defined by converging equations | null | null | null | null | math.GT math.CA math.MG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this note we investigate some applications of the Painlev\'e-Kuratowski convergence of closed sets in analysis that are motivated also by questions from singularity theory. First, we generalise to Lipschitz functions the classical theorem stating that given a sequence of smooth functions with locally uniformly convergent derivatives, we obtain the local uniform convergence of the functions themselves (provided they were convergent at one point). We use Clarke subdifferentials instead of derivative and Painlev\'e-Kuratowski convergence of their graphs instead of local uniform convergence. Next we focus on reverse theorem. We show that Painlev\'e-Kuratowski convergence of closed nonempty sets implies convergence of distance functions and Clark subdifferentials of squared distance functions, but does not imply convergence of Clark subdifferentials of distance functions. Finally we turn to the study of the behaviour of the fibres of a given function. We prove some general real counterparts of the Hurwitz theorem from complex analysis stating that the local uniform convergence of holomorphic functions implies the convergence of their sets of zeros. From the point of view of singularity theory our two theorems concern the convergence of the sets when their descriptions are convergent. They are also of interest in approximation theory. | [
{
"version": "v1",
"created": "Fri, 10 May 2024 08:36:15 GMT"
},
{
"version": "v2",
"created": "Sat, 07 Jun 2025 22:48:47 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Fatuła",
"Daniel",
""
]
] |
2405.06912 | Takeshi Torii | Takeshi Torii | A simple model for twisted arrow $\infty$-categories | 34 pages, major revision, sections 6 and 7 added | null | null | null | math.CT math.AT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Twisted arrow $\infty$-categories of $(\infty,1)$-categories were introduced by Lurie, and they have various applications in higher category theory. Abell\'{a}n Garc\'{i}a and Stern gave a generalization to twisted arrow $\infty$-categories of $(\infty,2)$-categories. In this paper we introduce another simple model for twisted arrow $\infty$-categories of $(\infty,2)$-categories. | [
{
"version": "v1",
"created": "Sat, 11 May 2024 04:55:20 GMT"
},
{
"version": "v2",
"created": "Fri, 06 Jun 2025 20:09:23 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Torii",
"Takeshi",
""
]
] |
2405.08485 | Minh N. Dao | Minh N. Dao, Tan Nhat Pham, and Phan Thanh Tung | Doubly relaxed forward-Douglas--Rachford splitting for the sum of two nonconvex and a DC function | null | Journal of Scientific Computing 104 (2025), no. 2, Paper No. 35 | 10.1007/s10915-025-02950-w | null | math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we consider a class of structured nonconvex nonsmooth optimization problems whose objective function is the sum of three nonconvex functions, one of which is expressed in a difference-of-convex (DC) form. This problem class covers several important structures in the literature including the sum of three functions and the general DC program. We propose a splitting algorithm and prove the subsequential convergence to a stationary point of the problem. The full sequential convergence, along with convergence rates for both the iterates and objective function values, is then established without requiring differentiability of the concave part. Our analysis not only extends but also unifies and improves recent convergence analyses in nonconvex settings. We benchmark our proposed algorithm with notable algorithms in the literature to show its competitiveness on a low rank matrix completion problem and a simutaneously sparse and low-rank matrix estimation problem. Our algorithm exhibits very competitive results compared to notable algorithms in the literature, on both synthetic data and public dataset. | [
{
"version": "v1",
"created": "Tue, 14 May 2024 10:18:38 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 11:07:19 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Dao",
"Minh N.",
""
],
[
"Pham",
"Tan Nhat",
""
],
[
"Tung",
"Phan Thanh",
""
]
] |
2405.08621 | Chen Feng | Tianhao Peng, Chen Feng, Duolikun Danier, Fan Zhang, Benoit Vallade, Alex Mackin, David Bull | RMT-BVQA: Recurrent Memory Transformer-based Blind Video Quality Assessment for Enhanced Video Content | This paper has been accepted by the ECCV 2024 AIM Advances in Image Manipulation workshop | null | 10.1007/978-3-031-91838-4_7 | null | eess.IV cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | With recent advances in deep learning, numerous algorithms have been developed to enhance video quality, reduce visual artifacts, and improve perceptual quality. However, little research has been reported on the quality assessment of enhanced content - the evaluation of enhancement methods is often based on quality metrics that were designed for compression applications. In this paper, we propose a novel blind deep video quality assessment (VQA) method specifically for enhanced video content. It employs a new Recurrent Memory Transformer (RMT) based network architecture to obtain video quality representations, which is optimized through a novel content-quality-aware contrastive learning strategy based on a new database containing 13K training patches with enhanced content. The extracted quality representations are then combined through linear regression to generate video-level quality indices. The proposed method, RMT-BVQA, has been evaluated on the VDPVE (VQA Dataset for Perceptual Video Enhancement) database through a five-fold cross validation. The results show its superior correlation performance when compared to ten existing no-reference quality metrics. | [
{
"version": "v1",
"created": "Tue, 14 May 2024 14:01:15 GMT"
},
{
"version": "v2",
"created": "Wed, 15 May 2024 09:35:56 GMT"
},
{
"version": "v3",
"created": "Wed, 04 Sep 2024 05:28:51 GMT"
},
{
"version": "v4",
"created": "Fri, 06 Sep 2024 01:05:22 GMT"
},
{
"version": "v5",
"created": "Thu, 10 Oct 2024 20:32:01 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Peng",
"Tianhao",
""
],
[
"Feng",
"Chen",
""
],
[
"Danier",
"Duolikun",
""
],
[
"Zhang",
"Fan",
""
],
[
"Vallade",
"Benoit",
""
],
[
"Mackin",
"Alex",
""
],
[
"Bull",
"David",
""
]
] |
2405.10144 | Vitor Araujo D | Vitor Araujo and Luciana Salgado | A characterization of physical measures for systems with mixed central behavior | 16 pages, 2 figure; improved results, references and extended comments | null | null | null | math.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that the existence of physical measures for $C^\infty$ smooth instances of certain partially hyperbolic dynamics, both continuous and discrete, exhibiting mixed behavior (positive and negative Lyapunov exponents) along the central non-uniformly hyperbolic multidimensional invariant direction, is equivalent to the existence of certain types of ``regular points'' on positive volume subsets, including Lyapunov regular points. This encompasses the $C^3$ robust class of multidimensional non-hyperbolic attractors obtained by Viana, and the $C^1$ robust classes of $3$-sectionally hyperbolic wild strange attractors presented by Shilnikov and Turaev, providing necessary and sufficient conditions for the existence of ergodic hyperbolic physical measures on these and other dynamical systems. | [
{
"version": "v1",
"created": "Thu, 16 May 2024 14:38:32 GMT"
},
{
"version": "v2",
"created": "Mon, 20 May 2024 13:36:41 GMT"
},
{
"version": "v3",
"created": "Fri, 06 Jun 2025 20:35:41 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Araujo",
"Vitor",
""
],
[
"Salgado",
"Luciana",
""
]
] |
2405.10178 | Robert Gaunt | Robert E. Gaunt, Saralees Nadarajah and Tibor K. Pog\'any | Infinite Divisibility of the Product of Two Correlated Normal Random Variables and Exact Distribution of the Sample Mean | 15 pages | null | null | null | math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We prove that the distribution of the product of two correlated normal random variables with arbitrary means and arbitrary variances is infinitely divisible. We also obtain exact formulas for the probability density function of the sum of independent copies of such random variables. | [
{
"version": "v1",
"created": "Thu, 16 May 2024 15:19:49 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Oct 2024 11:30:37 GMT"
},
{
"version": "v3",
"created": "Sun, 08 Jun 2025 16:58:52 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Gaunt",
"Robert E.",
""
],
[
"Nadarajah",
"Saralees",
""
],
[
"Pogány",
"Tibor K.",
""
]
] |
2405.13682 | Sho Sonoda Dr | Sho Sonoda, Yuka Hashimoto, Isao Ishikawa, Masahiro Ikeda | Deep Ridgelet Transform and Unified Universality Theorem for Deep and Shallow Joint-Group-Equivariant Machines | accepted at ICML2025 | null | null | null | cs.LG math.RT stat.ML | http://creativecommons.org/licenses/by/4.0/ | We present a constructive universal approximation theorem for learning machines equipped with joint-group-equivariant feature maps, called the joint-equivariant machines, based on the group representation theory. ``Constructive'' here indicates that the distribution of parameters is given in a closed-form expression known as the ridgelet transform. Joint-group-equivariance encompasses a broad class of feature maps that generalize classical group-equivariance. Particularly, fully-connected networks are not group-equivariant but are joint-group-equivariant. Our main theorem also unifies the universal approximation theorems for both shallow and deep networks. Until this study, the universality of deep networks has been shown in a different manner from the universality of shallow networks, but our results discuss them on common ground. Now we can understand the approximation schemes of various learning machines in a unified manner. As applications, we show the constructive universal approximation properties of four examples: depth-$n$ joint-equivariant machine, depth-$n$ fully-connected network, depth-$n$ group-convolutional network, and a new depth-$2$ network with quadratic forms whose universality has not been known. | [
{
"version": "v1",
"created": "Wed, 22 May 2024 14:25:02 GMT"
},
{
"version": "v2",
"created": "Thu, 03 Oct 2024 01:12:35 GMT"
},
{
"version": "v3",
"created": "Sun, 01 Dec 2024 16:49:16 GMT"
},
{
"version": "v4",
"created": "Fri, 31 Jan 2025 16:05:32 GMT"
},
{
"version": "v5",
"created": "Mon, 09 Jun 2025 06:38:02 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Sonoda",
"Sho",
""
],
[
"Hashimoto",
"Yuka",
""
],
[
"Ishikawa",
"Isao",
""
],
[
"Ikeda",
"Masahiro",
""
]
] |
2405.13762 | Krishna Somandepalli | Gwanghyun Kim, Alonso Martinez, Yu-Chuan Su, Brendan Jou, Jos\'e Lezama, Agrim Gupta, Lijun Yu, Lu Jiang, Aren Jansen, Jacob Walker and Krishna Somandepalli | A Versatile Diffusion Transformer with Mixture of Noise Levels for Audiovisual Generation | null | In Proceedings of the 38th Conference on Neural Information Processing Systems (NeurIPS 2024) | null | null | cs.CV cs.LG cs.MM cs.SD eess.AS | http://creativecommons.org/licenses/by/4.0/ | Training diffusion models for audiovisual sequences allows for a range of generation tasks by learning conditional distributions of various input-output combinations of the two modalities. Nevertheless, this strategy often requires training a separate model for each task which is expensive. Here, we propose a novel training approach to effectively learn arbitrary conditional distributions in the audiovisual space.Our key contribution lies in how we parameterize the diffusion timestep in the forward diffusion process. Instead of the standard fixed diffusion timestep, we propose applying variable diffusion timesteps across the temporal dimension and across modalities of the inputs. This formulation offers flexibility to introduce variable noise levels for various portions of the input, hence the term mixture of noise levels. We propose a transformer-based audiovisual latent diffusion model and show that it can be trained in a task-agnostic fashion using our approach to enable a variety of audiovisual generation tasks at inference time. Experiments demonstrate the versatility of our method in tackling cross-modal and multimodal interpolation tasks in the audiovisual space. Notably, our proposed approach surpasses baselines in generating temporally and perceptually consistent samples conditioned on the input. Project page: avdit2024.github.io | [
{
"version": "v1",
"created": "Wed, 22 May 2024 15:47:14 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Kim",
"Gwanghyun",
""
],
[
"Martinez",
"Alonso",
""
],
[
"Su",
"Yu-Chuan",
""
],
[
"Jou",
"Brendan",
""
],
[
"Lezama",
"José",
""
],
[
"Gupta",
"Agrim",
""
],
[
"Yu",
"Lijun",
""
],
[
"Jiang",
"Lu",
""
],
[
"Jansen",
"Aren",
""
],
[
"Walker",
"Jacob",
""
],
[
"Somandepalli",
"Krishna",
""
]
] |
2405.14019 | Alan Wang | Alan Q. Wang, Rachit Saluja, Heejong Kim, Xinzi He, Adrian Dalca, and Mert R. Sabuncu | BrainMorph: A Foundational Keypoint Model for Robust and Flexible Brain MRI Registration | Accepted for publication at the Journal of Machine Learning for Biomedical Imaging (MELBA) https://melba-journal.org/2025:010 | Machine.Learning.for.Biomedical.Imaging. 3 (2025) | 10.59275/j.melba.2025-59g7 | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | We present a keypoint-based foundation model for general purpose brain MRI registration, based on the recently-proposed KeyMorph framework. Our model, called BrainMorph, serves as a tool that supports multi-modal, pairwise, and scalable groupwise registration. BrainMorph is trained on a massive dataset of over 100,000 3D volumes, skull-stripped and non-skull-stripped, from nearly 16,000 unique healthy and diseased subjects. BrainMorph is robust to large misalignments, interpretable via interrogating automatically-extracted keypoints, and enables rapid and controllable generation of many plausible transformations with different alignment types and different degrees of nonlinearity at test-time. We demonstrate the superiority of BrainMorph in solving 3D rigid, affine, and nonlinear registration on a variety of multi-modal brain MRI scans of healthy and diseased subjects, in both the pairwise and groupwise setting. In particular, we show registration accuracy and speeds that surpass many classical and learning-based methods, especially in the context of large initial misalignments and large group settings. All code and models are available at https://github.com/alanqrwang/brainmorph. | [
{
"version": "v1",
"created": "Wed, 22 May 2024 21:52:30 GMT"
},
{
"version": "v2",
"created": "Fri, 24 May 2024 19:02:33 GMT"
},
{
"version": "v3",
"created": "Fri, 06 Jun 2025 18:23:04 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Wang",
"Alan Q.",
""
],
[
"Saluja",
"Rachit",
""
],
[
"Kim",
"Heejong",
""
],
[
"He",
"Xinzi",
""
],
[
"Dalca",
"Adrian",
""
],
[
"Sabuncu",
"Mert R.",
""
]
] |
2405.15049 | Bilas Paul | Bilas Paul, Shantanu Chakraborty, Ganga Sharma | Breaking Barriers: Investigating Gender Dynamics in Introductory Physics Lab Classes | 6 pages, 3 figures | Georgia Journal of Science 83, no. 2 (2025): 3 | null | null | physics.ed-ph | http://creativecommons.org/licenses/by/4.0/ | The persistent underrepresentation of women and other gender minorities in physical science fields has been an ongoing concern. This study investigates gender dynamics in introductory physics laboratory courses, focusing on whether students of different gender identities exhibit equal inclination and confidence in conducting lab experiments, and whether they face barriers that impact their participation. Conducted across three institutions and involving non-physics STEM students enrolled in algebra-based and calculus-based physics courses, the study found mixed results, with two institutions showing no significant gender-based differences in participation levels during lab activities, while one institution demonstrated significant differences. Chi-square tests revealed no significant association between gender and task preference or comfortability, though the small dataset suggests the need for further investigation. While quantitative analysis provided limited evidence of systematic barriers, qualitative feedback revealed that some female students experienced challenges related to gender dynamics, such as perceived assumptions about competence, being overlooked during discussions, and hesitation to voice opinions in male-dominated groups. These findings highlight the complex influence of gender and institutional factors on laboratory experiences and underscore the need for creating inclusive environments that promote equitable engagement and participation for all gender identities in STEM education. | [
{
"version": "v1",
"created": "Thu, 23 May 2024 20:56:26 GMT"
},
{
"version": "v2",
"created": "Wed, 25 Dec 2024 02:58:11 GMT"
},
{
"version": "v3",
"created": "Fri, 06 Jun 2025 22:55:35 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Paul",
"Bilas",
""
],
[
"Chakraborty",
"Shantanu",
""
],
[
"Sharma",
"Ganga",
""
]
] |
2405.15314 | Ilker Birbil | H\"useyin Tun\c{c}, Do\u{g}anay \"Ozese, \c{S}. \.Ilker Birbil, Donato Maragno, Marco Caserta, Mustafa Baydo\u{g}an | Output-Constrained Decision Trees | 23 pages, 3 figures | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Incorporating domain-specific constraints into machine learning models is essential for generating predictions that are both accurate and feasible in real-world applications. This paper introduces new methods for training Output-Constrained Regression Trees (OCRT), addressing the limitations of traditional decision trees in constrained multi-target regression tasks. We propose three approaches: M-OCRT, which uses split-based mixed integer programming to enforce constraints; E-OCRT, which employs an exhaustive search for optimal splits and solves constrained prediction problems at each decision node; and EP-OCRT, which applies post-hoc constrained optimization to tree predictions. To illustrate their potential uses in ensemble learning, we also introduce a random forest framework working under convex feasible sets. We validate the proposed methods through a computational study both on synthetic and industry-driven hierarchical time series datasets. Our results demonstrate that imposing constraints on decision tree training results in accurate and feasible predictions. | [
{
"version": "v1",
"created": "Fri, 24 May 2024 07:54:44 GMT"
},
{
"version": "v2",
"created": "Wed, 02 Oct 2024 10:05:38 GMT"
},
{
"version": "v3",
"created": "Mon, 09 Jun 2025 08:25:17 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Tunç",
"Hüseyin",
""
],
[
"Özese",
"Doğanay",
""
],
[
"Birbil",
"Ş. İlker",
""
],
[
"Maragno",
"Donato",
""
],
[
"Caserta",
"Marco",
""
],
[
"Baydoğan",
"Mustafa",
""
]
] |
2405.15386 | Stefan Diehl | Stefan Diehl, Kyungseon Joo, Kirill Semenov-Tian-Shansky, Christian Weiss, Vladimir Braun, Wen-Chen Chang, Pierre Chatagnon, Martha Constantinou, Yuxun Guo, Parada T. P. Hutauruk, Hyon-Suk Jo, Andrey Kim, Jun-Young Kim, Peter Kroll, Shunzo Kumano, Chang-Hwan Lee, Simonetta Liuti, Ronan McNulty, Hyeon-Dong Son, Pawel Sznajder, Ali Usman, Charlotte Van Hulse, Marc Vanderhaeghen and Michael Winn | Exploring Baryon Resonances with Transition Generalized Parton Distributions: Status and Perspectives | null | Eur. Phys. J. A 61, 131 (2025) | 10.1140/epja/s10050-025-01552-2 | JLAB-THY-24-4071 | hep-ph hep-ex nucl-th | http://creativecommons.org/licenses/by/4.0/ | QCD gives rise to a rich spectrum of excited baryon states. Understanding their internal structure is important for many areas of nuclear physics, such as nuclear forces, dense matter, and neutrino-nucleus interactions. Generalized parton distributions (GPDs) are an established tool for characterizing the QCD structure of the ground-state nucleon. They are used to create 3D tomographic images of the quark/gluon structure and quantify the mechanical properties such as the distribution of mass, angular momentum and forces in the system. Transition GPDs extend these concepts to $N \rightarrow N^\ast$ transitions and can be used to characterize the 3D structure and mechanical properties of baryon resonances. They can be probed in high-momentum-transfer exclusive electroproduction processes with resonance transitions $e + N \rightarrow e' + M + N^\ast$, such as deeply-virtual Compton scattering ($M = \gamma$) or meson production ($M = \pi, K$, $etc.$), and in related photon/hadron-induced processes. This White Paper describes a research program aiming to explore baryon resonance structure with transition GPDs. This includes the properties and interpretation of the transition GPDs, theoretical methods for structures and processes, first experimental results from JLab 12 GeV, future measurements with existing and planned facilities (JLab detector and energy upgrades, COMPASS/AMBER, EIC, EicC, J-PARC, LHC ultraperihperal collisions), and the theoretical and experimental developments needed to realize this program. | [
{
"version": "v1",
"created": "Fri, 24 May 2024 09:36:03 GMT"
},
{
"version": "v2",
"created": "Tue, 25 Mar 2025 12:52:00 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Diehl",
"Stefan",
""
],
[
"Joo",
"Kyungseon",
""
],
[
"Semenov-Tian-Shansky",
"Kirill",
""
],
[
"Weiss",
"Christian",
""
],
[
"Braun",
"Vladimir",
""
],
[
"Chang",
"Wen-Chen",
""
],
[
"Chatagnon",
"Pierre",
""
],
[
"Constantinou",
"Martha",
""
],
[
"Guo",
"Yuxun",
""
],
[
"Hutauruk",
"Parada T. P.",
""
],
[
"Jo",
"Hyon-Suk",
""
],
[
"Kim",
"Andrey",
""
],
[
"Kim",
"Jun-Young",
""
],
[
"Kroll",
"Peter",
""
],
[
"Kumano",
"Shunzo",
""
],
[
"Lee",
"Chang-Hwan",
""
],
[
"Liuti",
"Simonetta",
""
],
[
"McNulty",
"Ronan",
""
],
[
"Son",
"Hyeon-Dong",
""
],
[
"Sznajder",
"Pawel",
""
],
[
"Usman",
"Ali",
""
],
[
"Van Hulse",
"Charlotte",
""
],
[
"Vanderhaeghen",
"Marc",
""
],
[
"Winn",
"Michael",
""
]
] |
2405.15993 | Alberto Foss\`a | Alberto Foss\`a, Roberto Armellin, Emmanuel Delande, and Francesco Sanfedino | Efficient Multifidelity Uncertainty Propagation in the Presence of Process Noise | Submitted to the Journal of Guidance, Control, and Dynamics | null | 10.2514/1.G008519 | null | math.NA cs.NA | http://creativecommons.org/licenses/by-nc-nd/4.0/ | A multifidelity method for the nonlinear propagation of uncertainties in the presence of stochastic accelerations is presented. The proposed algorithm treats the uncertainty propagation (UP) problem by separating the propagation of the initial uncertainty from that of the process noise. The initial uncertainty is propagated using an adaptive Gaussian mixture model (GMM) method which exploits a low-fidelity dynamical model to minimize the computational costs. The effects of process noise are instead computed using the PoLynomial Algebra Stochastic Moments Analysis (PLASMA) technique, which considers a high-fidelity model of the stochastic dynamics. The main focus of the paper is on the latter and on the key idea to approximate the probability density function (pdf) of the solution by a polynomial representation of its moments, which are efficiently computed using differential algebra (DA) techniques. The two estimates are finally combined to restore the accuracy of the low-fidelity surrogate and account for both sources of uncertainty. The proposed approach is applied to the problem of nonlinear orbit UP and its performance compared to that of Monte Carlo (MC) simulations. | [
{
"version": "v1",
"created": "Sat, 25 May 2024 00:23:00 GMT"
},
{
"version": "v2",
"created": "Thu, 30 Jan 2025 00:37:23 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Fossà",
"Alberto",
""
],
[
"Armellin",
"Roberto",
""
],
[
"Delande",
"Emmanuel",
""
],
[
"Sanfedino",
"Francesco",
""
]
] |
2405.17351 | Yujie Wang | Yujie Wang, Praneeth Chakravarthula, Baoquan Chen | DOF-GS:Adjustable Depth-of-Field 3D Gaussian Splatting for Post-Capture Refocusing, Defocus Rendering and Blur Removal | Project page: https://dof-gs.github.io/ | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | 3D Gaussian Splatting (3DGS) techniques have recently enabled high-quality 3D scene reconstruction and real-time novel view synthesis. These approaches, however, are limited by the pinhole camera model and lack effective modeling of defocus effects. Departing from this, we introduce DOF-GS--a new 3DGS-based framework with a finite-aperture camera model and explicit, differentiable defocus rendering, enabling it to function as a post-capture control tool. By training with multi-view images with moderate defocus blur, DOF-GS learns inherent camera characteristics and reconstructs sharp details of the underlying scene, particularly, enabling rendering of varying DOF effects through on-demand aperture and focal distance control, post-capture and optimization. Additionally, our framework extracts circle-of-confusion cues during optimization to identify in-focus regions in input views, enhancing the reconstructed 3D scene details. Experimental results demonstrate that DOF-GS supports post-capture refocusing, adjustable defocus and high-quality all-in-focus rendering, from multi-view images with uncalibrated defocus blur. | [
{
"version": "v1",
"created": "Mon, 27 May 2024 16:54:49 GMT"
},
{
"version": "v2",
"created": "Thu, 27 Mar 2025 19:55:24 GMT"
},
{
"version": "v3",
"created": "Sat, 07 Jun 2025 05:21:24 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Wang",
"Yujie",
""
],
[
"Chakravarthula",
"Praneeth",
""
],
[
"Chen",
"Baoquan",
""
]
] |
2405.17656 | Najwa Laabid Mrs. | Najwa Laabid, Severi Rissanen, Markus Heinonen, Arno Solin, Vikas Garg | Equivariant Denoisers Cannot Copy Graphs: Align Your Graph Diffusion Models | Accepted at ICLR 2025. Code available at https://github.com/Aalto-QuML/DiffAlign | Proceedings of the International Conference on Learning Representations (ICLR) 2025 | null | null | cs.LG q-bio.QM | http://creativecommons.org/licenses/by/4.0/ | Graph diffusion models, dominant in graph generative modeling, remain underexplored for graph-to-graph translation tasks like chemical reaction prediction. We demonstrate that standard permutation equivariant denoisers face fundamental limitations in these tasks due to their inability to break symmetries in noisy inputs. To address this, we propose aligning input and target graphs to break input symmetries while preserving permutation equivariance in non-matching graph portions. Using retrosynthesis (i.e., the task of predicting precursors for synthesis of a given target molecule) as our application domain, we show how alignment dramatically improves discrete diffusion model performance from 5% to a SOTA-matching 54.7% top-1 accuracy. Code is available at https://github.com/Aalto-QuML/DiffAlign. | [
{
"version": "v1",
"created": "Mon, 27 May 2024 20:57:19 GMT"
},
{
"version": "v2",
"created": "Sun, 08 Jun 2025 13:22:06 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Laabid",
"Najwa",
""
],
[
"Rissanen",
"Severi",
""
],
[
"Heinonen",
"Markus",
""
],
[
"Solin",
"Arno",
""
],
[
"Garg",
"Vikas",
""
]
] |
2405.17813 | Owen Elliott | Owen Pendrigh Elliott, Jesse Clark | The Impacts of Data, Ordering, and Intrinsic Dimensionality on Recall in Hierarchical Navigable Small Worlds | 15 pages, 2 figures | null | 10.1145/3664190.3672512 | null | cs.IR | http://creativecommons.org/licenses/by/4.0/ | Vector search systems, pivotal in AI applications, often rely on the Hierarchical Navigable Small Worlds (HNSW) algorithm. However, the behaviour of HNSW under real-world scenarios using vectors generated with deep learning models remains under-explored. Existing Approximate Nearest Neighbours (ANN) benchmarks and research typically has an over-reliance on simplistic datasets like MNIST or SIFT1M and fail to reflect the complexity of current use-cases. Our investigation focuses on HNSW's efficacy across a spectrum of datasets, including synthetic vectors tailored to mimic specific intrinsic dimensionalities, widely-used retrieval benchmarks with popular embedding models, and proprietary e-commerce image data with CLIP models. We survey the most popular HNSW vector databases and collate their default parameters to provide a realistic fixed parameterisation for the duration of the paper.
We discover that the recall of approximate HNSW search, in comparison to exact K Nearest Neighbours (KNN) search, is linked to the vector space's intrinsic dimensionality and significantly influenced by the data insertion sequence. Our methodology highlights how insertion order, informed by measurable properties such as the pointwise Local Intrinsic Dimensionality (LID) or known categories, can shift recall by up to 12 percentage points. We also observe that running popular benchmark datasets with HNSW instead of KNN can shift rankings by up to three positions for some models. This work underscores the need for more nuanced benchmarks and design considerations in developing robust vector search systems using approximate vector search algorithms. This study presents a number of scenarios with varying real world applicability which aim to better increase understanding and future development of ANN algorithms and embedding | [
{
"version": "v1",
"created": "Tue, 28 May 2024 04:16:43 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Elliott",
"Owen Pendrigh",
""
],
[
"Clark",
"Jesse",
""
]
] |
2405.18186 | Cesare Donati | Cesare Donati, Martina Mammarella, Fabrizio Dabbene, Carlo Novara, Constantino Lagoa | Combining Off-White and Sparse Black Models in Multi-step Physics-based Systems Identification -- EXTENDED VERSION | null | In: Automatica, Volume 179, 2025, 112409 | 10.1016/j.automatica.2025.112409 | null | eess.SY cs.SY | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this paper, we propose a unified framework for identifying interpretable nonlinear dynamical models that preserve physical properties. The proposed approach integrates physical principles with black-box basis functions to compensate for unmodeled dynamics, ensuring accuracy over long prediction horizons and computational efficiency. Additionally, we introduce penalty terms to enforce physical consistency and stability during training. We provide a comprehensive analysis of theoretical properties related to multi-step nonlinear system identification, establishing bounds on parameter estimation errors and conditions for gradient stability and sparsity recovery. The proposed framework demonstrates significant potential for improving model accuracy and reliability in various engineering applications, making a substantial step towards the effective use of combined off-white and sparse black models in system identification. The effectiveness of the proposed approach is demonstrated on a nonlinear system identification benchmark. | [
{
"version": "v1",
"created": "Tue, 28 May 2024 13:54:48 GMT"
},
{
"version": "v2",
"created": "Tue, 06 Aug 2024 14:25:15 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Donati",
"Cesare",
""
],
[
"Mammarella",
"Martina",
""
],
[
"Dabbene",
"Fabrizio",
""
],
[
"Novara",
"Carlo",
""
],
[
"Lagoa",
"Constantino",
""
]
] |
2405.18380 | Pengxiang Li | Pengxiang Li, Lu Yin, Xiaowei Gao, Shiwei Liu | Outlier-weighed Layerwise Sampling for LLM Fine-tuning | null | null | null | null | cs.LG cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | The rapid advancements in Large Language Models (LLMs) have revolutionized various natural language processing tasks. However, the substantial size of LLMs presents significant challenges in training or fine-tuning. While parameter-efficient approaches such as low-rank adaptation (LoRA) have gained popularity, they often compromise performance compared to full-rank fine-tuning. In this paper, we propose Outlier-weighed Layerwise Sampling (OWS), a new memory-efficient fine-tuning approach, inspired by the layerwise outlier distribution of LLMs. Unlike LoRA, which adds extra adapters to all layers, OWS strategically assigns higher sampling probabilities to layers with more outliers, selectively sampling only a few layers and fine-tuning their pre-trained weights. To further increase the number of fine-tuned layers without a proportional rise in memory costs, we incorporate gradient low-rank projection, further boosting the approach's performance. Our extensive experiments across various architectures, including LLaMa2 and Mistral, demonstrate that OWS consistently outperforms baseline approaches, including full fine-tuning. Specifically, it achieves up to a 1.1% average accuracy gain on the Commonsense Reasoning benchmark, a 3.0% improvement on MMLU, and a notable 10% boost on MT-Bench, while being more memory efficient. OWS allows us to fine-tune 7B LLMs with only 21GB of memory. Our code is available at https://github.com/pixeli99/OWS. | [
{
"version": "v1",
"created": "Tue, 28 May 2024 17:22:22 GMT"
},
{
"version": "v2",
"created": "Sat, 12 Oct 2024 04:35:10 GMT"
},
{
"version": "v3",
"created": "Mon, 09 Jun 2025 03:55:25 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Li",
"Pengxiang",
""
],
[
"Yin",
"Lu",
""
],
[
"Gao",
"Xiaowei",
""
],
[
"Liu",
"Shiwei",
""
]
] |
2405.19289 | Avedis Neehus | Avedis Neehus, Frank Pollmann, Johannes Knolle | Genuine topological Anderson insulator from impurity induced chirality reversal | 7 pages, 2 figures, supplemental material: 7 pages, 5 figures | null | null | null | cond-mat.mes-hall cond-mat.dis-nn | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We investigate a model of Dirac fermions with Haldane type mass impurities which open a global topological gap even in the dilute limit. Surprisingly, we find that the chirality of this mass term, i.e., the sign of the Chern number, can be reversed by tuning the magnitude of the single-impurity scattering. Consequently, the disorder induces a phase disconnected from the clean topological phase, i.e., a genuine topological Anderson insulator. In seeming contradiction to the expectation that mass disorder is an irrelevant perturbation to the clean integer quantum Hall transition, the tri-critical point separating these two Chern insulating phases and a thermal metal phase is located at zero impurity density and connected to the appearance of a zero energy bound state in the continuum corresponding to a divergent Haldane mass impurity. Our conclusions based on the T-matrix expansion are substantiated by large scale Chebyshev-Polynomial-Green-Function numerics. We discuss possible experimental platforms. | [
{
"version": "v1",
"created": "Wed, 29 May 2024 17:18:51 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 14:57:01 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Neehus",
"Avedis",
""
],
[
"Pollmann",
"Frank",
""
],
[
"Knolle",
"Johannes",
""
]
] |
2405.19507 | Zachary Ferguson | Maxine Perroni-Scharf, Zachary Ferguson, Thomas Butrille, Carlos Portela, Mina Konakovi\'c Lukovi\'c | Data-Efficient Discovery of Hyperelastic TPMS Metamaterials with Extreme Energy Dissipation | null | In SIGGRAPH Conference Papers 2025, August 10--14, 2025, Vancouver, BC, Canada. ACM, New York, NY, USA, 12 pages | 10.1145/3721238.3730759 | null | cs.GR | http://creativecommons.org/licenses/by/4.0/ | Triply periodic minimal surfaces (TPMS) are a class of metamaterials with a variety of applications and well-known primitive morphologies. We present a new method for discovering novel microscale TPMS structures with exceptional energy-dissipation capabilities, achieving double the energy absorption of the best existing TPMS primitive structure. Our approach employs a parametric representation, allowing seamless interpolation between structures and representing a rich TPMS design space. As simulations are intractable for efficiently optimizing microscale hyperelastic structures, we propose a sample-efficient computational strategy for rapid discovery with limited empirical data from 3D-printed and tested samples that ensures high-fidelity results. We achieve this by leveraging a predictive uncertainty-aware Deep Ensembles model to identify which structures to fabricate and test next. We iteratively refine our model through batch Bayesian optimization, selecting structures for fabrication that maximize exploration of the performance space and exploitation of our energy-dissipation objective. Using our method, we produce the first open-source dataset of hyperelastic microscale TPMS structures, including a set of novel structures that demonstrate extreme energy dissipation capabilities, and show several potential applications of these structures. | [
{
"version": "v1",
"created": "Wed, 29 May 2024 20:43:25 GMT"
},
{
"version": "v2",
"created": "Sat, 07 Jun 2025 03:21:36 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Perroni-Scharf",
"Maxine",
""
],
[
"Ferguson",
"Zachary",
""
],
[
"Butrille",
"Thomas",
""
],
[
"Portela",
"Carlos",
""
],
[
"Luković",
"Mina Konaković",
""
]
] |
2405.20771 | Jingwei Li | Jingwei Li, Jing Dong, Tianxing He, Jingzhao Zhang | Towards Black-Box Membership Inference Attack for Diffusion Models | null | null | null | null | cs.CR cs.AI cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Given the rising popularity of AI-generated art and the associated copyright concerns, identifying whether an artwork was used to train a diffusion model is an important research topic. The work approaches this problem from the membership inference attack (MIA) perspective. We first identify the limitation of applying existing MIA methods for proprietary diffusion models: the required access of internal U-nets. To address the above problem, we introduce a novel membership inference attack method that uses only the image-to-image variation API and operates without access to the model's internal U-net. Our method is based on the intuition that the model can more easily obtain an unbiased noise prediction estimate for images from the training set. By applying the API multiple times to the target image, averaging the outputs, and comparing the result to the original image, our approach can classify whether a sample was part of the training set. We validate our method using DDIM and Stable Diffusion setups and further extend both our approach and existing algorithms to the Diffusion Transformer architecture. Our experimental results consistently outperform previous methods. | [
{
"version": "v1",
"created": "Sat, 25 May 2024 12:47:58 GMT"
},
{
"version": "v2",
"created": "Tue, 26 Nov 2024 05:05:32 GMT"
},
{
"version": "v3",
"created": "Wed, 27 Nov 2024 03:48:21 GMT"
},
{
"version": "v4",
"created": "Sat, 07 Jun 2025 08:23:08 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Li",
"Jingwei",
""
],
[
"Dong",
"Jing",
""
],
[
"He",
"Tianxing",
""
],
[
"Zhang",
"Jingzhao",
""
]
] |
2405.20907 | Zoe Nieraeth | Zoe Nieraeth | The Muckenhoupt condition | 45 pages, incorporated referee comments, elaborated on the main results in the introduction, proved an equivalence of two of the stated conjectures, added a section on Musielak-Orlicz spaces | null | null | null | math.FA math.CA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The goal of this paper is to unify the theory of weights beyond the setting of weighted Lebesgue spaces in the general setting of quasi-Banach function spaces. We prove new characterizations for the boundedness of singular integrals, pose several conjectures, and prove partial results related to the duality of the Hardy-Littlewood maximal operator. Furthermore, we give an overview of the theory applied to weighted variable Lebesgue, Morrey, and Musielak-Orlicz spaces. | [
{
"version": "v1",
"created": "Fri, 31 May 2024 15:18:08 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 17:35:30 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Nieraeth",
"Zoe",
""
]
] |
2406.00032 | Ying Zhang | Ying Zhang, Xiaofeng Li, Zhaoyang Liu, Haipeng Zhang | Paths of A Million People: Extracting Life Trajectories from Wikipedia | Accepted to ICWSM 2025. 15 pages | null | 10.1609/icwsm.v19i1.35930 | null | cs.CL cs.AI cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The life trajectories of notable people have been studied to pinpoint the times and places of significant events such as birth, death, education, marriage, competition, work, speeches, scientific discoveries, artistic achievements, and battles. Understanding how these individuals interact with others provides valuable insights for broader research into human dynamics. However, the scarcity of trajectory data in terms of volume, density, and inter-person interactions, limits relevant studies from being comprehensive and interactive. We mine millions of biography pages from Wikipedia and tackle the generalization problem stemming from the variety and heterogeneity of the trajectory descriptions. Our ensemble model COSMOS, which combines the idea of semi-supervised learning and contrastive learning, achieves an F1 score of 85.95%. For this task, we also create a hand-curated dataset, WikiLifeTrajectory, consisting of 8,852 (person, time, location) triplets as ground truth. Besides, we perform an empirical analysis on the trajectories of 8,272 historians to demonstrate the validity of the extracted results. To facilitate the research on trajectory extractions and help the analytical studies to construct grand narratives, we make our code, the million-level extracted trajectories, and the WikiLifeTrajectory dataset publicly available. | [
{
"version": "v1",
"created": "Sat, 25 May 2024 06:57:33 GMT"
},
{
"version": "v2",
"created": "Sun, 21 Jul 2024 06:52:40 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Zhang",
"Ying",
""
],
[
"Li",
"Xiaofeng",
""
],
[
"Liu",
"Zhaoyang",
""
],
[
"Zhang",
"Haipeng",
""
]
] |
2406.00391 | Yen-Nhi Nguyen Ngoc | Nhi Ngoc-Yen Nguyen, Le-Huy Tu, Dieu-Phuong Nguyen, Nhat-Tan Do, Minh Triet Thai, Bao-Thien Nguyen-Tat | DS@BioMed at ImageCLEFmedical Caption 2024: Enhanced Attention Mechanisms in Medical Caption Generation through Concept Detection Integration | null | DS@BioMed at ImageCLEF 2024, CEUR Workshop Proceedings, Vol. 3740. https://ceur-ws.org/Vol-3740/paper-157.pdf | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Purpose: Our study presents an enhanced approach to medical image caption generation by integrating concept detection into attention mechanisms. Method: This method utilizes sophisticated models to identify critical concepts within medical images, which are then refined and incorporated into the caption generation process. Results: Our concept detection task, which employed the Swin-V2 model, achieved an F1 score of 0.58944 on the validation set and 0.61998 on the private test set, securing the third position. For the caption prediction task, our BEiT+BioBart model, enhanced with concept integration and post-processing techniques, attained a BERTScore of 0.60589 on the validation set and 0.5794 on the private test set, placing ninth. Conclusion: These results underscore the efficacy of concept-aware algorithms in generating precise and contextually appropriate medical descriptions. The findings demonstrate that our approach significantly improves the quality of medical image captions, highlighting its potential to enhance medical image interpretation and documentation, thereby contributing to improved healthcare outcomes. | [
{
"version": "v1",
"created": "Sat, 01 Jun 2024 10:14:33 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Nguyen",
"Nhi Ngoc-Yen",
""
],
[
"Tu",
"Le-Huy",
""
],
[
"Nguyen",
"Dieu-Phuong",
""
],
[
"Do",
"Nhat-Tan",
""
],
[
"Thai",
"Minh Triet",
""
],
[
"Nguyen-Tat",
"Bao-Thien",
""
]
] |
2406.00736 | Jasson Vindas | Jasson Vindas | On the estimate $M(x)=o(x)$ for Beurling generalized numbers | 9 pages | Analysis Math. 50 (2024), 1131-1140 | 10.1007/s10476-024-00061-6 | null | math.NT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that the sum function of the M\"{o}bius function of a Beurling number system must satisfy the asymptotic bound $M(x)=o(x)$ if it satisfies the prime number theorem and its prime distribution function arises from a monotone perturbation of either the classical prime numbers or the logarithmic integral. | [
{
"version": "v1",
"created": "Sun, 02 Jun 2024 13:13:06 GMT"
},
{
"version": "v2",
"created": "Sun, 22 Sep 2024 19:38:43 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Vindas",
"Jasson",
""
]
] |
2406.00971 | Vahid Azizi | Vahid Azizi, Fatemeh Koochaki | MiniGPT-Reverse-Designing: Predicting Image Adjustments Utilizing MiniGPT-4 | 8 pages, 7 figures | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Vision-Language Models (VLMs) have recently seen significant advancements through integrating with Large Language Models (LLMs). The VLMs, which process image and text modalities simultaneously, have demonstrated the ability to learn and understand the interaction between images and texts across various multi-modal tasks. Reverse designing, which could be defined as a complex vision-language task, aims to predict the edits and their parameters, given a source image, an edited version, and an optional high-level textual edit description. This task requires VLMs to comprehend the interplay between the source image, the edited version, and the optional textual context simultaneously, going beyond traditional vision-language tasks. In this paper, we extend and fine-tune MiniGPT-4 for the reverse designing task. Our experiments demonstrate the extensibility of off-the-shelf VLMs, specifically MiniGPT-4, for more complex tasks such as reverse designing. Code is available at this \href{https://github.com/VahidAz/MiniGPT-Reverse-Designing} | [
{
"version": "v1",
"created": "Mon, 03 Jun 2024 03:59:29 GMT"
},
{
"version": "v2",
"created": "Fri, 30 Aug 2024 01:50:37 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Azizi",
"Vahid",
""
],
[
"Koochaki",
"Fatemeh",
""
]
] |
2406.01440 | Vishwanath Shukla | Pinaki Dutta, Sayan Choudhury, Vishwanath Shukla | Prethermalization in the PXP Model under Continuous Quasiperiodic Driving | 14 pages, 6 figures | Phys. Rev. B 111, 064303 (2025) | 10.1103/PhysRevB.111.064303 | null | cond-mat.quant-gas cond-mat.stat-mech cond-mat.str-el quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Motivated by recent experiments realizing long-lived non-equilibrium states in aperiodically driven quantum many-body systems, we investigate the dynamics of a quasiperiodically driven Rydberg atom chain in the strong Rydberg blockage regime. In this regime, the system is kinetically constrained and the `PXP' model describes its dynamics. Even without driving, the PXP model exhibits many-body scarring and resultant persistent oscillations for dynamics originating from the N\'{e}el-ordered initial state. We demonstrate that a rich array of dynamical behaviors emerge when the system is subjected to a continuous drive. In the high-frequency regime, the system exhibits revivals and oscillations for the N\'{e}el ordered initial state both for periodic and quasi-periodic drives. We trace the origin of this non-ergodicity to an effective PXP Hamiltonian for both of these driving protocols in this regime. Furthermore, we demonstrate that the behavior of the fidelity and the entanglement entropy is non-monotonic at low frequencies in the high-amplitude regime. This leads to several re-entrant scarring transitions both for both the N\'{e}el-ordered and the fully polarized initial state. Our results demonstrate that continuous quasi-periodic drive protocols can provide a promising route to realize prethermal phases of matter in kinetically constrained systems. | [
{
"version": "v1",
"created": "Mon, 03 Jun 2024 15:30:02 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Dutta",
"Pinaki",
""
],
[
"Choudhury",
"Sayan",
""
],
[
"Shukla",
"Vishwanath",
""
]
] |
2406.01493 | Jiahao Shao | Jiahao Shao, Yuanbo Yang, Hongyu Zhou, Youmin Zhang, Yujun Shen, Vitor Guizilini, Yue Wang, Matteo Poggi, Yiyi Liao | Learning Temporally Consistent Video Depth from Video Diffusion Priors | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | This work addresses the challenge of streamed video depth estimation, which expects not only per-frame accuracy but, more importantly, cross-frame consistency. We argue that sharing contextual information between frames or clips is pivotal in fostering temporal consistency. Therefore, we reformulate depth prediction into a conditional generation problem to provide contextual information within a clip and across clips. Specifically, we propose a consistent context-aware training and inference strategy for arbitrarily long videos to provide cross-clip context. We sample independent noise levels for each frame within a clip during training while using a sliding window strategy and initializing overlapping frames with previously predicted frames without adding noise. Moreover, we design an effective training strategy to provide context within a clip. Extensive experimental results validate our design choices and demonstrate the superiority of our approach, dubbed ChronoDepth. Project page: https://xdimlab.github.io/ChronoDepth/. | [
{
"version": "v1",
"created": "Mon, 03 Jun 2024 16:20:24 GMT"
},
{
"version": "v2",
"created": "Tue, 04 Jun 2024 03:33:52 GMT"
},
{
"version": "v3",
"created": "Mon, 02 Dec 2024 17:10:34 GMT"
},
{
"version": "v4",
"created": "Sat, 07 Jun 2025 07:24:16 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Shao",
"Jiahao",
""
],
[
"Yang",
"Yuanbo",
""
],
[
"Zhou",
"Hongyu",
""
],
[
"Zhang",
"Youmin",
""
],
[
"Shen",
"Yujun",
""
],
[
"Guizilini",
"Vitor",
""
],
[
"Wang",
"Yue",
""
],
[
"Poggi",
"Matteo",
""
],
[
"Liao",
"Yiyi",
""
]
] |
2406.02998 | Alberto Guandalini | Alberto Guandalini, Ryosuke Senga, Yung-Chang Lin, Kazu Suenaga, Paolo Barone, Francesco Mauri, Thomas Pichler, Christian Kramberger | Direct observation of the vanishing EELS cross section in graphene | null | null | null | null | cond-mat.mes-hall cond-mat.other | http://creativecommons.org/licenses/by/4.0/ | In transmission electron energy-loss spectroscopy, the cross section in 2D is quenched by kinematic effects once the momentum transfer becomes smaller than a critical value set by $q_z$, the momentum loss parallel to the beam. Our highly momentum ($\Delta q = 0.02$~\r{A}$^{-1}$) and energy ($\Delta E = 45$~meV) resolved setup is instrumental on delivering the unprecedented experimental verification of quenched 2D EEL spectra on freestanding graphene at momentum transfers $q$ below $0.06$\r{A}$^{-1}$. We retrieve the intrinsic uniform dielectric response of graphene from measured spectra by quantifying the kinematic suppression. | [
{
"version": "v1",
"created": "Wed, 05 Jun 2024 06:57:29 GMT"
},
{
"version": "v2",
"created": "Fri, 31 Jan 2025 10:45:29 GMT"
},
{
"version": "v3",
"created": "Mon, 09 Jun 2025 09:16:54 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Guandalini",
"Alberto",
""
],
[
"Senga",
"Ryosuke",
""
],
[
"Lin",
"Yung-Chang",
""
],
[
"Suenaga",
"Kazu",
""
],
[
"Barone",
"Paolo",
""
],
[
"Mauri",
"Francesco",
""
],
[
"Pichler",
"Thomas",
""
],
[
"Kramberger",
"Christian",
""
]
] |
2406.04233 | Bernardo Leite | Bernardo Leite, Tom\'as Freitas Os\'orio, Henrique Lopes Cardoso | FairytaleQA Translated: Enabling Educational Question and Answer Generation in Less-Resourced Languages | Preprint - Accepted for publication at ECTEL 2024 | EC-TEL 2024, Lecture Notes in Computer Science, vol. 15159, pp. 222-236, Springer, 2024 | 10.1007/978-3-031-72315-5_16 | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Question Answering (QA) datasets are crucial in assessing reading comprehension skills for both machines and humans. While numerous datasets have been developed in English for this purpose, a noticeable void exists in less-resourced languages. To alleviate this gap, our paper introduces machine-translated versions of FairytaleQA, a renowned QA dataset designed to assess and enhance narrative comprehension skills in young children. By employing fine-tuned, modest-scale models, we establish benchmarks for both Question Generation (QG) and QA tasks within the translated datasets. In addition, we present a case study proposing a model for generating question-answer pairs, with an evaluation incorporating quality metrics such as question well-formedness, answerability, relevance, and children suitability. Our evaluation prioritizes quantifying and describing error cases, along with providing directions for future work. This paper contributes to the advancement of QA and QG research in less-resourced languages, promoting accessibility and inclusivity in the development of these models for reading comprehension. The code and data is publicly available at github.com/bernardoleite/fairytaleqa-translated. | [
{
"version": "v1",
"created": "Thu, 06 Jun 2024 16:31:47 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Jun 2024 15:39:17 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Leite",
"Bernardo",
""
],
[
"Osório",
"Tomás Freitas",
""
],
[
"Cardoso",
"Henrique Lopes",
""
]
] |
2406.04548 | Hsiao-Ying Lu | Hsiao-Ying Lu, Yiran Li, Ujwal Pratap Krishna Kaluvakolanu Thyagarajan, Kwan-Liu Ma | GNNAnatomy: Rethinking Model-Level Explanations for Graph Neural Networks | null | null | null | null | cs.LG cs.IR cs.SI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Graph Neural Networks (GNNs) achieve outstanding performance across graph-based tasks but remain difficult to interpret. In this paper, we revisit foundational assumptions underlying model-level explanation methods for GNNs, namely: (1) maximizing classification confidence yields representative explanations, (2) a single explanation suffices for an entire class of graphs, and (3) explanations are inherently trustworthy. We identify pitfalls resulting from these assumptions: methods that optimize for classification confidence may overlook partially learned patterns; topological diversity across graph subsets within the same class is often underrepresented; and explanations alone offer limited support for building user trust when applied to new datasets or models. This paper introduces GNNAnatomy, a distillation-based method designed to generate explanations while avoiding these pitfalls. GNNAnatomy first characterizes graph topology using graphlets, a set of fundamental substructures. We then train a transparent multilayer perceptron surrogate to directly approximate GNN predictions based on the graphlet representations. By analyzing the weights assigned to each graphlet, we identify the most discriminative topologies, which serve as GNN explanations. To account for structural diversity within a class, GNNAnatomy generates explanations at the required granularity through an interface that supports human-AI teaming. This interface helps users identify subsets of graphs where distinct critical substructures drive class differentiation, enabling multi-grained explanations. Additionally, by enabling exploration and linking explanations back to input graphs, the interface fosters greater transparency and trust. We evaluate GNNAnatomy on both synthetic and real-world datasets through quantitative metrics and qualitative comparisons with state-of-the-art model-level explainable GNN methods. | [
{
"version": "v1",
"created": "Thu, 06 Jun 2024 23:09:54 GMT"
},
{
"version": "v2",
"created": "Mon, 23 Sep 2024 17:38:54 GMT"
},
{
"version": "v3",
"created": "Fri, 06 Jun 2025 21:23:31 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Lu",
"Hsiao-Ying",
""
],
[
"Li",
"Yiran",
""
],
[
"Thyagarajan",
"Ujwal Pratap Krishna Kaluvakolanu",
""
],
[
"Ma",
"Kwan-Liu",
""
]
] |
2406.04905 | Caterina Stoppato | Steven G. Krantz, Marco M. Peloso, Caterina Stoppato | On a higher-dimensional worm domain and its geometric properties | 31 pages, 2 figures | J. Lond. Math. Soc., 111(6):e70195 (2025) | 10.1112/jlms.70195 | null | math.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We construct new $3$-dimensional variants of the classical Diederich-Fornaess worm domain. We show that they are smoothly bounded, pseudoconvex, and have nontrivial Nebenh\"{u}lle. We also show that their Bergman projections do not preserve the Sobolev space for sufficiently large Sobolev indices. | [
{
"version": "v1",
"created": "Fri, 07 Jun 2024 12:58:07 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 17:47:24 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Krantz",
"Steven G.",
""
],
[
"Peloso",
"Marco M.",
""
],
[
"Stoppato",
"Caterina",
""
]
] |
2406.05225 | Zhiyang Wang | Zhiyang Wang, Juan Cervino, Alejandro Ribeiro | A Manifold Perspective on the Statistical Generalization of Graph Neural Networks | 38 pages,22 figures | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Graph Neural Networks (GNNs) extend convolutional neural networks to operate on graphs. Despite their impressive performances in various graph learning tasks, the theoretical understanding of their generalization capability is still lacking. Previous GNN generalization bounds ignore the underlying graph structures, often leading to bounds that increase with the number of nodes -- a behavior contrary to the one experienced in practice. In this paper, we take a manifold perspective to establish the statistical generalization theory of GNNs on graphs sampled from a manifold in the spectral domain. As demonstrated empirically, we prove that the generalization bounds of GNNs decrease linearly with the size of the graphs in the logarithmic scale, and increase linearly with the spectral continuity constants of the filter functions. Notably, our theory explains both node-level and graph-level tasks. Our result has two implications: i) guaranteeing the generalization of GNNs to unseen data over manifolds; ii) providing insights into the practical design of GNNs, i.e., restrictions on the discriminability of GNNs are necessary to obtain a better generalization performance. We demonstrate our generalization bounds of GNNs using synthetic and multiple real-world datasets. | [
{
"version": "v1",
"created": "Fri, 07 Jun 2024 19:25:02 GMT"
},
{
"version": "v2",
"created": "Tue, 20 Aug 2024 18:19:39 GMT"
},
{
"version": "v3",
"created": "Tue, 10 Sep 2024 16:16:13 GMT"
},
{
"version": "v4",
"created": "Thu, 07 Nov 2024 19:49:10 GMT"
},
{
"version": "v5",
"created": "Tue, 12 Nov 2024 15:05:00 GMT"
},
{
"version": "v6",
"created": "Fri, 06 Jun 2025 22:52:34 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Wang",
"Zhiyang",
""
],
[
"Cervino",
"Juan",
""
],
[
"Ribeiro",
"Alejandro",
""
]
] |
2406.06777 | Khiem Le | Khiem Le, Zhichun Guo, Kaiwen Dong, Xiaobao Huang, Bozhao Nan, Roshni Iyer, Xiangliang Zhang, Olaf Wiest, Wei Wang, Ting Hua, Nitesh V. Chawla | MolX: Enhancing Large Language Models for Molecular Understanding With A Multi-Modal Extension | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large Language Models (LLMs) with their strong task-handling capabilities have shown remarkable advancements across a spectrum of fields, moving beyond natural language understanding. However, their proficiency within the chemistry domain remains restricted, especially in solving professional molecule-related tasks. This challenge is attributed to their inherent limitations in comprehending molecules using only common textual representations, i.e., SMILES strings. In this study, we seek to enhance the ability of LLMs to comprehend molecules by equipping them with a multi-modal external module, namely MolX. In particular, instead of directly using a SMILES string to represent a molecule, we utilize specific encoders to extract fine-grained features from both SMILES string and 2D molecular graph representations for feeding into an LLM. Moreover, a handcrafted molecular fingerprint is incorporated to leverage its embedded domain knowledge. Then, to establish an alignment between MolX and the LLM's textual input space, the whole model in which the LLM is frozen, is pre-trained with a versatile strategy including a diverse set of tasks. Experimental evaluations show that our proposed method outperforms baselines across 4 downstream molecule-related tasks ranging from molecule-to-text translation to retrosynthesis, with and without fine-tuning the LLM, while only introducing a small number of trainable parameters -- 0.53\% and 0.82\%, respectively. | [
{
"version": "v1",
"created": "Mon, 10 Jun 2024 20:25:18 GMT"
},
{
"version": "v2",
"created": "Thu, 13 Jun 2024 03:54:49 GMT"
},
{
"version": "v3",
"created": "Fri, 28 Jun 2024 03:07:29 GMT"
},
{
"version": "v4",
"created": "Thu, 22 Aug 2024 02:06:31 GMT"
},
{
"version": "v5",
"created": "Wed, 02 Apr 2025 22:20:34 GMT"
},
{
"version": "v6",
"created": "Sun, 08 Jun 2025 10:34:54 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Le",
"Khiem",
""
],
[
"Guo",
"Zhichun",
""
],
[
"Dong",
"Kaiwen",
""
],
[
"Huang",
"Xiaobao",
""
],
[
"Nan",
"Bozhao",
""
],
[
"Iyer",
"Roshni",
""
],
[
"Zhang",
"Xiangliang",
""
],
[
"Wiest",
"Olaf",
""
],
[
"Wang",
"Wei",
""
],
[
"Hua",
"Ting",
""
],
[
"Chawla",
"Nitesh V.",
""
]
] |
2406.09043 | Soumadeep Saha | Soumadeep Saha and Sutanoya Chakraborty and Saptarshi Saha and Utpal Garain | Language Models are Crossword Solvers | 10 pages, 3 figures, 6 Appendix. Accepted to NAACL 2025 | NAACL 2025 (Volume 1: Long Papers), pages 2074-2090 | 10.18653/v1/2025.naacl-long.104 | 2025.naacl-long.104 | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Crosswords are a form of word puzzle that require a solver to demonstrate a high degree of proficiency in natural language understanding, wordplay, reasoning, and world knowledge, along with adherence to character and length constraints. In this paper we tackle the challenge of solving crosswords with large language models (LLMs). We demonstrate that the current generation of language models shows significant competence at deciphering cryptic crossword clues and outperforms previously reported state-of-the-art (SoTA) results by a factor of 2-3 in relevant benchmarks. We also develop a search algorithm that builds off this performance to tackle the problem of solving full crossword grids with out-of-the-box LLMs for the very first time, achieving an accuracy of 93% on New York Times crossword puzzles. Additionally, we demonstrate that LLMs generalize well and are capable of supporting answers with sound rationale. | [
{
"version": "v1",
"created": "Thu, 13 Jun 2024 12:29:27 GMT"
},
{
"version": "v2",
"created": "Fri, 14 Jun 2024 21:29:40 GMT"
},
{
"version": "v3",
"created": "Sun, 09 Feb 2025 14:26:45 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Saha",
"Soumadeep",
""
],
[
"Chakraborty",
"Sutanoya",
""
],
[
"Saha",
"Saptarshi",
""
],
[
"Garain",
"Utpal",
""
]
] |
2406.09096 | Venkat Abhignan Y | Venkat Abhignan | Casimir energy for $N$ constant conductivity $\delta$-plates with a neural network perception | 16 pages, 10 figures, added new discussion with neural networks | Eur. Phys. J. Plus 140, 491 (2025) | 10.1140/epjp/s13360-025-06436-4 | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | The Casimir energy for $N$ $\delta$-function plates depends on the multiple scattering parameter $\Delta$. This $N$-body interaction was distributed into interactions with nearest neighbour scattering and next-to-nearest neighbour scattering based on partitions of $N-1$ and its permutations. Implementing this methodology, we investigate the Casimir interaction for multiple plates with constant conductivity relatable to Graphene. We also study the Casimir energy between a perfect magnetic conductor and multiple constant conductivity $\delta$ plates, which results in Boyer repulsion. In the asymptotic limit for ideal boundary conditions, the results become simple where the multiple scattering parameter $\Delta$ consists only of the nearest neighbour scattering term. Further, we used neural networks to analyze the Casimir energy in the Boyer repulsion configurations to understand the influence of pairwise energies on the many-body energy. The neural network could distinguish the repulsive and attractive forces depending on the regions of varying conductivity. | [
{
"version": "v1",
"created": "Thu, 13 Jun 2024 13:26:14 GMT"
},
{
"version": "v2",
"created": "Fri, 06 Jun 2025 20:34:40 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Abhignan",
"Venkat",
""
]
] |
2406.09401 | Tai Wang | Ruiyuan Lyu, Jingli Lin, Tai Wang, Shuai Yang, Xiaohan Mao, Yilun Chen, Runsen Xu, Haifeng Huang, Chenming Zhu, Dahua Lin, Jiangmiao Pang | MMScan: A Multi-Modal 3D Scene Dataset with Hierarchical Grounded Language Annotations | Follow-up of EmbodiedScan (camera-ready version). A multi-modal 3D dataset with the most-ever comprehensive language annotations for 3D-LLMs. Project page: https://tai-wang.github.io/mmscan/ | null | null | null | cs.CV cs.AI cs.RO | http://creativecommons.org/licenses/by-nc-sa/4.0/ | With the emergence of LLMs and their integration with other data modalities, multi-modal 3D perception attracts more attention due to its connectivity to the physical world and makes rapid progress. However, limited by existing datasets, previous works mainly focus on understanding object properties or inter-object spatial relationships in a 3D scene. To tackle this problem, this paper builds the first largest ever multi-modal 3D scene dataset and benchmark with hierarchical grounded language annotations, MMScan. It is constructed based on a top-down logic, from region to object level, from a single target to inter-target relationships, covering holistic aspects of spatial and attribute understanding. The overall pipeline incorporates powerful VLMs via carefully designed prompts to initialize the annotations efficiently and further involve humans' correction in the loop to ensure the annotations are natural, correct, and comprehensive. Built upon existing 3D scanning data, the resulting multi-modal 3D dataset encompasses 1.4M meta-annotated captions on 109k objects and 7.7k regions as well as over 3.04M diverse samples for 3D visual grounding and question-answering benchmarks. We evaluate representative baselines on our benchmarks, analyze their capabilities in different aspects, and showcase the key problems to be addressed in the future. Furthermore, we use this high-quality dataset to train state-of-the-art 3D visual grounding and LLMs and obtain remarkable performance improvement both on existing benchmarks and in-the-wild evaluation. Codes, datasets, and benchmarks will be available at https://github.com/OpenRobotLab/EmbodiedScan. | [
{
"version": "v1",
"created": "Thu, 13 Jun 2024 17:59:30 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 07:53:16 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Lyu",
"Ruiyuan",
""
],
[
"Lin",
"Jingli",
""
],
[
"Wang",
"Tai",
""
],
[
"Yang",
"Shuai",
""
],
[
"Mao",
"Xiaohan",
""
],
[
"Chen",
"Yilun",
""
],
[
"Xu",
"Runsen",
""
],
[
"Huang",
"Haifeng",
""
],
[
"Zhu",
"Chenming",
""
],
[
"Lin",
"Dahua",
""
],
[
"Pang",
"Jiangmiao",
""
]
] |
2406.10281 | Yan Sun | Patrick Chao, Yan Sun, Edgar Dobriban, Hamed Hassani | Watermarking Language Models with Error Correcting Codes | null | null | null | null | cs.CR cs.CL cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent progress in large language models enables the creation of realistic machine-generated content. Watermarking is a promising approach to distinguish machine-generated text from human text, embedding statistical signals in the output that are ideally undetectable to humans. We propose a watermarking framework that encodes such signals through an error correcting code. Our method, termed robust binary code (RBC) watermark, introduces no noticeable degradation in quality. We evaluate our watermark on base and instruction fine-tuned models and find that our watermark is robust to edits, deletions, and translations. We provide an information-theoretic perspective on watermarking, a powerful statistical test for detection and for generating $p$-values, and theoretical guarantees. Our empirical findings suggest our watermark is fast, powerful, and robust, comparing favorably to the state-of-the-art. | [
{
"version": "v1",
"created": "Wed, 12 Jun 2024 05:13:09 GMT"
},
{
"version": "v2",
"created": "Wed, 12 Feb 2025 02:11:10 GMT"
},
{
"version": "v3",
"created": "Sun, 18 May 2025 05:31:35 GMT"
},
{
"version": "v4",
"created": "Sun, 08 Jun 2025 23:08:39 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Chao",
"Patrick",
""
],
[
"Sun",
"Yan",
""
],
[
"Dobriban",
"Edgar",
""
],
[
"Hassani",
"Hamed",
""
]
] |
2406.11011 | Jiachen T. Wang | Jiachen T. Wang, Prateek Mittal, Dawn Song, Ruoxi Jia | Data Shapley in One Training Run | ICLR 2025 Outstanding Paper Runner-Up | null | null | null | cs.LG cs.CL stat.ML | http://creativecommons.org/licenses/by/4.0/ | Data Shapley provides a principled framework for attributing data's contribution within machine learning contexts. However, existing approaches require re-training models on different data subsets, which is computationally intensive, foreclosing their application to large-scale models. Furthermore, they produce the same attribution score for any models produced by running the learning algorithm, meaning they cannot perform targeted attribution towards a specific model obtained from a single run of the algorithm. This paper introduces In-Run Data Shapley, which addresses these limitations by offering scalable data attribution for a target model of interest. In its most efficient implementation, our technique incurs negligible additional runtime compared to standard model training. This dramatic efficiency improvement makes it possible to perform data attribution for the foundation model pretraining stage for the first time. We present several case studies that offer fresh insights into pretraining data's contribution and discuss their implications for copyright in generative AI and pretraining data curation. | [
{
"version": "v1",
"created": "Sun, 16 Jun 2024 17:09:24 GMT"
},
{
"version": "v2",
"created": "Sat, 29 Jun 2024 23:05:32 GMT"
},
{
"version": "v3",
"created": "Sat, 07 Jun 2025 15:16:12 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Wang",
"Jiachen T.",
""
],
[
"Mittal",
"Prateek",
""
],
[
"Song",
"Dawn",
""
],
[
"Jia",
"Ruoxi",
""
]
] |
2406.11458 | Nir Rosenfeld | Maayan Ehrenberg, Roy Ganz, Nir Rosenfeld | Adversaries With Incentives: A Strategic Alternative to Adversarial Robustness | null | ICLR 2025 | null | null | cs.LG cs.GT | http://creativecommons.org/licenses/by/4.0/ | Adversarial training aims to defend against adversaries: malicious opponents whose sole aim is to harm predictive performance in any way possible. This presents a rather harsh perspective, which we assert results in unnecessarily conservative training. As an alternative, we propose to model opponents as simply pursuing their own goals--rather than working directly against the classifier. Employing tools from strategic modeling, our approach enables knowledge or beliefs regarding the opponent's possible incentives to be used as inductive bias for learning. Accordingly, our method of strategic training is designed to defend against all opponents within an 'incentive uncertainty set'. This resorts to adversarial learning when the set is maximal, but offers potential gains when the set can be appropriately reduced. We conduct a series of experiments that show how even mild knowledge regarding the opponent's incentives can be useful, and that the degree of potential gains depends on how these incentives relate to the structure of the learning task. | [
{
"version": "v1",
"created": "Mon, 17 Jun 2024 12:20:59 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Feb 2025 18:14:27 GMT"
},
{
"version": "v3",
"created": "Mon, 09 Jun 2025 14:28:26 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Ehrenberg",
"Maayan",
""
],
[
"Ganz",
"Roy",
""
],
[
"Rosenfeld",
"Nir",
""
]
] |
2406.11682 | Shangqing Tu | Shangqing Tu, Zhuoran Pan, Wenxuan Wang, Zhexin Zhang, Yuliang Sun, Jifan Yu, Hongning Wang, Lei Hou, Juanzi Li | Knowledge-to-Jailbreak: Investigating Knowledge-driven Jailbreaking Attacks for Large Language Models | Accepted by KDD 2025 research track | null | null | null | cs.CL cs.AI cs.CR | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Large language models (LLMs) have been increasingly applied to various domains, which triggers increasing concerns about LLMs' safety on specialized domains, e.g. medicine. Despite prior explorations on general jailbreaking attacks, there are two challenges for applying existing attacks on testing the domain-specific safety of LLMs: (1) Lack of professional knowledge-driven attacks, (2) Insufficient coverage of domain knowledge. To bridge this gap, we propose a new task, knowledge-to-jailbreak, which aims to generate jailbreaking attacks from domain knowledge, requiring both attack effectiveness and knowledge relevance. We collect a large-scale dataset with 12,974 knowledge-jailbreak pairs and fine-tune a large language model as jailbreak-generator, to produce domain knowledge-specific jailbreaks. Experiments on 13 domains and 8 target LLMs demonstrate the effectiveness of jailbreak-generator in generating jailbreaks that are both threatening to the target LLMs and relevant to the given knowledge. We also apply our method to an out-of-domain knowledge base, showing that jailbreak-generator can generate jailbreaks that are comparable in harmfulness to those crafted by human experts. Data and code are available at: https://github.com/THU-KEG/Knowledge-to-Jailbreak/. | [
{
"version": "v1",
"created": "Mon, 17 Jun 2024 15:59:59 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 02:18:21 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Tu",
"Shangqing",
""
],
[
"Pan",
"Zhuoran",
""
],
[
"Wang",
"Wenxuan",
""
],
[
"Zhang",
"Zhexin",
""
],
[
"Sun",
"Yuliang",
""
],
[
"Yu",
"Jifan",
""
],
[
"Wang",
"Hongning",
""
],
[
"Hou",
"Lei",
""
],
[
"Li",
"Juanzi",
""
]
] |
2406.12091 | Pankayaraj Pathmanathan | Pankayaraj Pathmanathan, Souradip Chakraborty, Xiangyu Liu, Yongyuan Liang, Furong Huang | Is poisoning a real threat to LLM alignment? Maybe more so than you think | null | AAAI 2025 | null | null | cs.LG cs.CL cs.CR | http://creativecommons.org/licenses/by/4.0/ | Recent advancements in Reinforcement Learning with Human Feedback (RLHF) have significantly impacted the alignment of Large Language Models (LLMs). The sensitivity of reinforcement learning algorithms such as Proximal Policy Optimization (PPO) has led to new line work on Direct Policy Optimization (DPO), which treats RLHF in a supervised learning framework. The increased practical use of these RLHF methods warrants an analysis of their vulnerabilities. In this work, we investigate the vulnerabilities of DPO to poisoning attacks under different scenarios and compare the effectiveness of preference poisoning, a first of its kind. We comprehensively analyze DPO's vulnerabilities under different types of attacks, i.e., backdoor and non-backdoor attacks, and different poisoning methods across a wide array of language models, i.e., LLama 7B, Mistral 7B, and Gemma 7B. We find that unlike PPO-based methods, which, when it comes to backdoor attacks, require at least 4\% of the data to be poisoned to elicit harmful behavior, we exploit the true vulnerabilities of DPO more simply so we can poison the model with only as much as 0.5\% of the data. We further investigate the potential reasons behind the vulnerability and how well this vulnerability translates into backdoor vs non-backdoor attacks. | [
{
"version": "v1",
"created": "Mon, 17 Jun 2024 21:06:00 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Jun 2024 17:56:17 GMT"
},
{
"version": "v3",
"created": "Tue, 04 Feb 2025 15:57:59 GMT"
},
{
"version": "v4",
"created": "Mon, 09 Jun 2025 14:00:06 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Pathmanathan",
"Pankayaraj",
""
],
[
"Chakraborty",
"Souradip",
""
],
[
"Liu",
"Xiangyu",
""
],
[
"Liang",
"Yongyuan",
""
],
[
"Huang",
"Furong",
""
]
] |
2406.12481 | Hamidreza Mohammadi | Saeed Akbari-Kourbolagh, Ali Mahdifar, Hamidreza Mohammadi | Curvature effects on stimulated parametric down-conversion process: an analog model | 14 pages, 8 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | In this paper, we utilize an analog model of the general relativity and investigate the influence of spatial curvature on quantum properties of stimulated parametric down-conversion process. For this purpose, we use two-mode sphere coherent state as the input beams of the aforementioned process. These states are realization of coherent states of two-dimensional harmonic oscillator, which lies on a two-dimension sphere. We calculate the entanglement of output states of stimulated parametric down conversion process, measured by linear entropy, and show that it depends on the spatial curvature. So, by preparing the suitable two-mode sphere coherent states, it is possible to control the entanglement between the output states in the laboratory. In addition, we consider mean number and Mandel parameter of the output states of the process and also, their cross-correlation function, as the convince measures of non-classical behaviors. | [
{
"version": "v1",
"created": "Tue, 18 Jun 2024 10:36:48 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 15:31:37 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Akbari-Kourbolagh",
"Saeed",
""
],
[
"Mahdifar",
"Ali",
""
],
[
"Mohammadi",
"Hamidreza",
""
]
] |
2406.12782 | Ramon Gonzalez Rodriguez | Jos\'e Manuel Fern\'andez Vilaboa, Ram\'on Gonz\'alez Rodr\'iguez and Brais Ramos P\'erez | Relative Rota-Baxter operators, modules and projections | arXiv admin note: text overlap with arXiv:2404.12231 | null | 10.21136/CMJ.2025.0467-24 | null | math.RA | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The present article is devoted to introduce, in a braided monoidal setting, the notion of module over a relative Rota-Baxter operator. It is proved that there exists an adjunction between the category of modules associated to an invertible relative Rota-Baxter operator and the category of modules associated to a Hopf brace, which induces an equivalence by assuming certain additional hypothesis. Moreover, the notion of projection between relative Rota-Baxter operators is defined, and it is proved that those which are called ``strong'' give rise to a module according to the previous definition in the cocommutative setting. | [
{
"version": "v1",
"created": "Tue, 18 Jun 2024 16:49:20 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Vilaboa",
"José Manuel Fernández",
""
],
[
"Rodríguez",
"Ramón González",
""
],
[
"Pérez",
"Brais Ramos",
""
]
] |
2406.13342 | Hwiyeol Jo | Hwiyeol Jo, Hyunwoo Lee, Kang Min Yoo, Taiwoo Park | ZeroDL: Zero-shot Distribution Learning for Text Clustering via Large Language Models | Accepted at ACL2025(Findings) | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | The advancements in large language models (LLMs) have brought significant progress in NLP tasks. However, if a task cannot be fully described in prompts, the models could fail to carry out the task. In this paper, we propose a simple yet effective method to contextualize a task toward a LLM. The method utilizes (1) open-ended zero-shot inference from the entire dataset, (2) aggregate the inference results, and (3) finally incorporate the aggregated meta-information for the actual task. We show the effectiveness in text clustering tasks, empowering LLMs to perform text-to-text-based clustering and leading to improvements on several datasets. Furthermore, we explore the generated class labels for clustering, showing how the LLM understands the task through data. | [
{
"version": "v1",
"created": "Wed, 19 Jun 2024 08:48:05 GMT"
},
{
"version": "v2",
"created": "Sat, 07 Jun 2025 13:44:07 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Jo",
"Hwiyeol",
""
],
[
"Lee",
"Hyunwoo",
""
],
[
"Yoo",
"Kang Min",
""
],
[
"Park",
"Taiwoo",
""
]
] |
2406.13544 | Yuchang Zhu | Yuchang Zhu, Jintang Li, Yatao Bian, Zibin Zheng, Liang Chen | One Fits All: Learning Fair Graph Neural Networks for Various Sensitive Attributes | Published in KDD 2024 | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent studies have highlighted fairness issues in Graph Neural Networks (GNNs), where they produce discriminatory predictions against specific protected groups categorized by sensitive attributes such as race and age. While various efforts to enhance GNN fairness have made significant progress, these approaches are often tailored to specific sensitive attributes. Consequently, they necessitate retraining the model from scratch to accommodate changes in the sensitive attribute requirement, resulting in high computational costs. To gain deeper insights into this issue, we approach the graph fairness problem from a causal modeling perspective, where we identify the confounding effect induced by the sensitive attribute as the underlying reason. Motivated by this observation, we formulate the fairness problem in graphs from an invariant learning perspective, which aims to learn invariant representations across environments. Accordingly, we propose a graph fairness framework based on invariant learning, namely FairINV, which enables the training of fair GNNs to accommodate various sensitive attributes within a single training session. Specifically, FairINV incorporates sensitive attribute partition and trains fair GNNs by eliminating spurious correlations between the label and various sensitive attributes. Experimental results on several real-world datasets demonstrate that FairINV significantly outperforms state-of-the-art fairness approaches, underscoring its effectiveness. Our code is available via: https://github.com/ZzoomD/FairINV/. | [
{
"version": "v1",
"created": "Wed, 19 Jun 2024 13:30:17 GMT"
},
{
"version": "v2",
"created": "Wed, 03 Jul 2024 02:53:47 GMT"
},
{
"version": "v3",
"created": "Mon, 09 Jun 2025 05:47:07 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Zhu",
"Yuchang",
""
],
[
"Li",
"Jintang",
""
],
[
"Bian",
"Yatao",
""
],
[
"Zheng",
"Zibin",
""
],
[
"Chen",
"Liang",
""
]
] |
2406.13725 | Trang Pham | Viet-Hoang Tran, Trang Pham, Tho Tran, Minh Khoi Nguyen Nhat, Thanh Chu, Tam Le, Tan M. Nguyen | Tree-Sliced Wasserstein Distance: A Geometric Perspective | Accepted to ICML 2025 | null | null | null | cs.LG cs.AI stat.ML | http://creativecommons.org/licenses/by/4.0/ | Many variants of Optimal Transport (OT) have been developed to address its heavy computation. Among them, notably, Sliced Wasserstein (SW) is widely used for application domains by projecting the OT problem onto one-dimensional lines, and leveraging the closed-form expression of the univariate OT to reduce the computational burden. However, projecting measures onto low-dimensional spaces can lead to a loss of topological information. To mitigate this issue, in this work, we propose to replace one-dimensional lines with a more intricate structure, called tree systems. This structure is metrizable by a tree metric, which yields a closed-form expression for OT problems on tree systems. We provide an extensive theoretical analysis to formally define tree systems with their topological properties, introduce the concept of splitting maps, which operate as the projection mechanism onto these structures, then finally propose a novel variant of Radon transform for tree systems and verify its injectivity. This framework leads to an efficient metric between measures, termed Tree-Sliced Wasserstein distance on Systems of Lines (TSW-SL). By conducting a variety of experiments on gradient flows, image style transfer, and generative models, we illustrate that our proposed approach performs favorably compared to SW and its variants. | [
{
"version": "v1",
"created": "Wed, 19 Jun 2024 17:40:11 GMT"
},
{
"version": "v2",
"created": "Fri, 02 May 2025 04:45:34 GMT"
},
{
"version": "v3",
"created": "Mon, 09 Jun 2025 09:52:00 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Tran",
"Viet-Hoang",
""
],
[
"Pham",
"Trang",
""
],
[
"Tran",
"Tho",
""
],
[
"Nhat",
"Minh Khoi Nguyen",
""
],
[
"Chu",
"Thanh",
""
],
[
"Le",
"Tam",
""
],
[
"Nguyen",
"Tan M.",
""
]
] |
2406.13960 | Yi Cheng | Yi Cheng, Wenge Liu, Kaishuai Xu, Wenjun Hou, Yi Ouyang, Chak Tou Leong, Wenjie Li, Xian Wu, Yefeng Zheng | AutoPal: Autonomous Adaptation to Users for Personal AI Companionship | null | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Previous research has demonstrated the potential of AI agents to act as companions that can provide constant emotional support for humans. In this paper, we emphasize the necessity of autonomous adaptation in personal AI companionship, an underexplored yet promising direction. Such adaptability is crucial as it can facilitate more tailored interactions with users and allow the agent to evolve in response to users' changing needs. However, imbuing agents with autonomous adaptability presents unique challenges, including identifying optimal adaptations to meet users' expectations and ensuring a smooth transition during the adaptation process. To address them, we devise a hierarchical framework, AutoPal, that enables controllable and authentic adjustments to the agent's persona based on user interactions. A personamatching dataset is constructed to facilitate the learning of optimal persona adaptations. Extensive experiments demonstrate the effectiveness of AutoPal and highlight the importance of autonomous adaptability in AI companionship. | [
{
"version": "v1",
"created": "Thu, 20 Jun 2024 03:02:38 GMT"
},
{
"version": "v2",
"created": "Wed, 02 Oct 2024 13:59:40 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Oct 2024 03:10:13 GMT"
},
{
"version": "v4",
"created": "Sat, 07 Jun 2025 15:11:04 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Cheng",
"Yi",
""
],
[
"Liu",
"Wenge",
""
],
[
"Xu",
"Kaishuai",
""
],
[
"Hou",
"Wenjun",
""
],
[
"Ouyang",
"Yi",
""
],
[
"Leong",
"Chak Tou",
""
],
[
"Li",
"Wenjie",
""
],
[
"Wu",
"Xian",
""
],
[
"Zheng",
"Yefeng",
""
]
] |
2406.14572 | Vlad Vinogradov | Vlad Vinogradov, Ivan Izmailov, Simon Steshin, Kong T. Nguyen | Bioptic B1: A Target-Agnostic Potency-Based Small Molecules Search Engine | null | null | null | null | q-bio.QM cs.AI cs.IR | http://creativecommons.org/licenses/by/4.0/ | Recent successes in virtual screening have been made possible by large models and extensive chemical libraries. However, combining these elements is challenging: the larger the model, the more expensive it is to run, making ultra-large libraries unfeasible. To address this, we developed a target-agnostic, efficacy-based molecule search model, which allows us to find structurally dissimilar molecules with similar biological activities. We used the best practices to design fast retrieval system, based on processor-optimized SIMD instructions, enabling us to screen the ultra-large 40B Enamine REAL library with 100\% recall rate. We extensively benchmarked our model and several state-of-the-art models for both speed performance and retrieval quality of novel molecules. | [
{
"version": "v1",
"created": "Thu, 13 Jun 2024 17:53:29 GMT"
},
{
"version": "v2",
"created": "Thu, 27 Jun 2024 18:00:31 GMT"
},
{
"version": "v3",
"created": "Mon, 01 Jul 2024 01:33:10 GMT"
},
{
"version": "v4",
"created": "Fri, 06 Jun 2025 22:20:55 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Vinogradov",
"Vlad",
""
],
[
"Izmailov",
"Ivan",
""
],
[
"Steshin",
"Simon",
""
],
[
"Nguyen",
"Kong T.",
""
]
] |
2406.15577 | Jiaozi Wang | Jiaozi Wang and Philipp Strasberg | Decoherence of Histories: Chaotic Versus Integrable Systems | 14 pages, 22 figures | Phys. Rev. Lett. 134, 220401 (2025) | 10.1103/m8vq-l449 | null | cond-mat.stat-mech gr-qc quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the emergence of decoherent histories in isolated systems based on exact numerical integration of the Schr\"odinger equation for a Heisenberg chain. We reveal that the nature of the system, which we switch from (i) chaotic to (ii) interacting integrable to (iii) non-interacting integrable, strongly impacts decoherence \new{of coarse spin observables}. From a finite size scaling law we infer a strong exponential suppression of coherences for (i), a weak exponential suppression for (ii) and no exponential suppression for (iii) on a relevant short (nonequilibrium) time scale. Moreover, for longer times we find stronger decoherence for (i) but the opposite for (ii), hinting even at a possible power-law decay for (ii) at equilibrium time scales. This behaviour is encoded in the multi-time properties of the quantum histories and it can not be explained by environmentally induced decoherence. Our results suggest that chaoticity plays a crucial role in the emergence of classicality in finite size systems. | [
{
"version": "v1",
"created": "Fri, 21 Jun 2024 18:25:12 GMT"
},
{
"version": "v2",
"created": "Tue, 28 Jan 2025 08:46:00 GMT"
},
{
"version": "v3",
"created": "Thu, 15 May 2025 12:24:34 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Wang",
"Jiaozi",
""
],
[
"Strasberg",
"Philipp",
""
]
] |
2406.16087 | Chen Wang | Chen Wang, Kaiyi Ji, Junyi Geng, Zhongqiang Ren, Taimeng Fu, Fan Yang, Yifan Guo, Haonan He, Xiangyu Chen, Zitong Zhan, Qiwei Du, Shaoshu Su, Bowen Li, Yuheng Qiu, Yi Du, Qihang Li, Yifan Yang, Xiao Lin, Zhipeng Zhao | Imperative Learning: A Self-supervised Neuro-Symbolic Learning Framework for Robot Autonomy | null | The International Journal of Robotics Research (IJRR), 2025 | null | null | cs.RO cs.AI cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Data-driven methods such as reinforcement and imitation learning have achieved remarkable success in robot autonomy. However, their data-centric nature still hinders them from generalizing well to ever-changing environments. Moreover, labeling data for robotic tasks is often impractical and expensive. To overcome these challenges, we introduce a new self-supervised neuro-symbolic (NeSy) computational framework, imperative learning (IL), for robot autonomy, leveraging the generalization abilities of symbolic reasoning. The framework of IL consists of three primary components: a neural module, a reasoning engine, and a memory system. We formulate IL as a special bilevel optimization (BLO), which enables reciprocal learning over the three modules. This overcomes the label-intensive obstacles associated with data-driven approaches and takes advantage of symbolic reasoning concerning logical reasoning, physical principles, geometric analysis, etc. We discuss several optimization techniques for IL and verify their effectiveness in five distinct robot autonomy tasks including path planning, rule induction, optimal control, visual odometry, and multi-robot routing. Through various experiments, we show that IL can significantly enhance robot autonomy capabilities and we anticipate that it will catalyze further research across diverse domains. | [
{
"version": "v1",
"created": "Sun, 23 Jun 2024 12:02:17 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Jul 2024 03:20:26 GMT"
},
{
"version": "v3",
"created": "Thu, 25 Jul 2024 07:50:58 GMT"
},
{
"version": "v4",
"created": "Wed, 07 Aug 2024 02:36:19 GMT"
},
{
"version": "v5",
"created": "Sat, 25 Jan 2025 04:11:34 GMT"
},
{
"version": "v6",
"created": "Sun, 08 Jun 2025 18:29:46 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Wang",
"Chen",
""
],
[
"Ji",
"Kaiyi",
""
],
[
"Geng",
"Junyi",
""
],
[
"Ren",
"Zhongqiang",
""
],
[
"Fu",
"Taimeng",
""
],
[
"Yang",
"Fan",
""
],
[
"Guo",
"Yifan",
""
],
[
"He",
"Haonan",
""
],
[
"Chen",
"Xiangyu",
""
],
[
"Zhan",
"Zitong",
""
],
[
"Du",
"Qiwei",
""
],
[
"Su",
"Shaoshu",
""
],
[
"Li",
"Bowen",
""
],
[
"Qiu",
"Yuheng",
""
],
[
"Du",
"Yi",
""
],
[
"Li",
"Qihang",
""
],
[
"Yang",
"Yifan",
""
],
[
"Lin",
"Xiao",
""
],
[
"Zhao",
"Zhipeng",
""
]
] |
2406.16267 | Ye Han | Ye Han, James Fishbaugh, Jared Vicory, Silvani Amin, Matthew Daemer, Hannah E. Dewey, Yan Wang, Analise M. Sulentic, Alana Cianciulli, Andras Lasso, Matthew A. Jolley, Beatriz Paniagua | Geodesic-based Predictive Shape Modeling of the Right Ventricle in Patients with Hypoplastic Left Heart Syndrome | 14 pages, 3 figures, 1 table | null | null | null | physics.med-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Hypoplastic left heart syndrome (HLHS) is characterized by severe underdevelopment of left ventricle requiring staged surgical reconstruction (stages) to allow the right ventricle (RV) alone to support the circulation. In this setting changes in RV size and shape over time reflect adaptations to single-ventricle physiology, dysfunction of the associated tricuspid valve (TV), and are associated with circulatory failure. As such, an accurate prediction of the RV shape of a patient would inform understanding of both RV and TV failure, as well as clinical prognosis and associated decision making. We present a geodesic-based predictive shape modeling framework applied a cohort of RVs obtained from 15 HLHS patients at three individual time points. Reasonable predictions on stage 1 RV shapes can generated using pre-stage 1 RV shapes and two predictors from prior clinical and demographic measures. Our results demonstrate the future potential for a data-driven method to predict how the morphology of the RV of an individual patient will change in size and shape over time. Future studies will seek to expand the training sample size and integrate more comprehensive demographic and morphological data into the proposed predictive model. | [
{
"version": "v1",
"created": "Mon, 24 Jun 2024 02:21:04 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 16:51:54 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Han",
"Ye",
""
],
[
"Fishbaugh",
"James",
""
],
[
"Vicory",
"Jared",
""
],
[
"Amin",
"Silvani",
""
],
[
"Daemer",
"Matthew",
""
],
[
"Dewey",
"Hannah E.",
""
],
[
"Wang",
"Yan",
""
],
[
"Sulentic",
"Analise M.",
""
],
[
"Cianciulli",
"Alana",
""
],
[
"Lasso",
"Andras",
""
],
[
"Jolley",
"Matthew A.",
""
],
[
"Paniagua",
"Beatriz",
""
]
] |
2406.16627 | Minqiang Xu | Liang Chen, Minqiang Xu and Haizhang Zhang | A Random Integration Algorithm for High-dimensional Function Spaces | null | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a novel random integration algorithm that boasts both high convergence order and polynomial tractability for functions characterized by sparse frequencies or rapidly decaying Fourier coefficients. Specifically, for integration in periodic isotropic Sobolev space and the isotropic Sobolev space with compact support, our approach attains a nearly optimal root mean square error (RMSE) bound. In contrast to previous nearly optimal algorithms, our method exhibits polynomial tractability, ensuring that the number of samples does not scale exponentially with increasing dimensions. Our integration algorithm also enjoys nearly optimal bound for weighted Korobov space. Furthermore, the algorithm can be applied without the need for prior knowledge of weights, distinguishing it from the component-by-component algorithm. For integration in the Wiener algebra, the sample complexity of our algorithm is independent of the decay rate of Fourier coefficients. The effectiveness of the integration is confirmed through numerical experiments. | [
{
"version": "v1",
"created": "Mon, 24 Jun 2024 13:21:58 GMT"
},
{
"version": "v2",
"created": "Sat, 18 Jan 2025 10:01:42 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Apr 2025 15:05:16 GMT"
},
{
"version": "v4",
"created": "Sun, 08 Jun 2025 14:41:41 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Chen",
"Liang",
""
],
[
"Xu",
"Minqiang",
""
],
[
"Zhang",
"Haizhang",
""
]
] |
2406.17458 | Sebastian Hafner | Sebastian Hafner, Heng Fang, Hossein Azizpour, Yifang Ban | Continuous Urban Change Detection from Satellite Image Time Series with Temporal Feature Refinement and Multi-Task Integration | Accepted to IEEE Transactions on Geoscience and Remote Sensing, Code will be available at https://github.com/SebastianHafner/ContUrbanCD.git | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Urbanization advances at unprecedented rates, leading to negative environmental and societal impacts. Remote sensing can help mitigate these effects by supporting sustainable development strategies with accurate information on urban growth. Deep learning-based methods have achieved promising urban change detection results from optical satellite image pairs using convolutional neural networks (ConvNets), transformers, and a multi-task learning setup. However, bi-temporal methods are limited for continuous urban change detection, i.e., the detection of changes in consecutive image pairs of satellite image time series (SITS), as they fail to fully exploit multi-temporal data (> 2 images). Existing multi-temporal change detection methods, on the other hand, collapse the temporal dimension, restricting their ability to capture continuous urban changes. Additionally, multi-task learning methods lack integration approaches that combine change and segmentation outputs. To address these challenges, we propose a continuous urban change detection framework incorporating two key modules. The temporal feature refinement (TFR) module employs self-attention to improve ConvNet-based multi-temporal building representations. The temporal dimension is preserved in the TFR module, enabling the detection of continuous changes. The multi-task integration (MTI) module utilizes Markov networks to find an optimal building map time series based on segmentation and dense change outputs. The proposed framework effectively identifies urban changes based on high-resolution SITS acquired by the PlanetScope constellation (F1 score 0.551), Gaofen-2 (F1 score 0.440), and WorldView-2 (F1 score 0.543). Moreover, our experiments on three challenging datasets demonstrate the effectiveness of the proposed framework compared to bi-temporal and multi-temporal urban change detection and segmentation methods. | [
{
"version": "v1",
"created": "Tue, 25 Jun 2024 10:53:57 GMT"
},
{
"version": "v2",
"created": "Fri, 17 Jan 2025 10:59:58 GMT"
},
{
"version": "v3",
"created": "Mon, 09 Jun 2025 09:32:53 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Hafner",
"Sebastian",
""
],
[
"Fang",
"Heng",
""
],
[
"Azizpour",
"Hossein",
""
],
[
"Ban",
"Yifang",
""
]
] |
2406.17470 | Jintao Yan | Jintao Yan, Tan Chen, Yuxuan Sun, Zhaojun Nan, Sheng Zhou, Zhisheng Niu | Dynamic Scheduling for Vehicle-to-Vehicle Communications Enhanced Federated Learning | Accepted by the IEEE Transactions on Wireless Communications | null | null | null | cs.LG cs.AI cs.DC cs.IT math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Leveraging the computing and sensing capabilities of vehicles, vehicular federated learning (VFL) has been applied to edge training for connected vehicles. The dynamic and interconnected nature of vehicular networks presents unique opportunities to harness direct vehicle-to-vehicle (V2V) communications, enhancing VFL training efficiency. In this paper, we formulate a stochastic optimization problem to optimize the VFL training performance, considering the energy constraints and mobility of vehicles, and propose a V2V-enhanced dynamic scheduling (VEDS) algorithm to solve it. The model aggregation requirements of VFL and the limited transmission time due to mobility result in a stepwise objective function, which presents challenges in solving the problem. We thus propose a derivative-based drift-plus-penalty method to convert the long-term stochastic optimization problem to an online mixed integer nonlinear programming (MINLP) problem, and provide a theoretical analysis to bound the performance gap between the online solution and the offline optimal solution. Further analysis of the scheduling priority reduces the original problem into a set of convex optimization problems, which are efficiently solved using the interior-point method. Experimental results demonstrate that compared with the state-of-the-art benchmarks, the proposed algorithm enhances the image classification accuracy on the CIFAR-10 dataset by 4.20% and reduces the average displacement errors on the Argoverse trajectory prediction dataset by 9.82%. | [
{
"version": "v1",
"created": "Tue, 25 Jun 2024 11:15:53 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 00:05:40 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Yan",
"Jintao",
""
],
[
"Chen",
"Tan",
""
],
[
"Sun",
"Yuxuan",
""
],
[
"Nan",
"Zhaojun",
""
],
[
"Zhou",
"Sheng",
""
],
[
"Niu",
"Zhisheng",
""
]
] |
2406.18415 | Luis Crespo | Luis Crespo, \'Alvaro Pelayo | The $p$-adic Jaynes-Cummings model in symplectic geometry | 58 pages, 14 figures | J. Nonlinear Sci. 35, 66 (2025) | 10.1007/s00332-025-10161-8 | null | math.SG math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The notion of classical $p$-adic integrable system on a $p$-adic symplectic manifold was proposed by Voevodsky, Warren and the second author a decade ago in analogy with the real case. In the present paper we introduce and study, from the viewpoint of symplectic geometry and topology, the basic properties of the $p$-adic version of the classical Jaynes-Cummings model. The Jaynes-Cummings model is a fundamental example of integrable system going back to the work of Jaynes and Cummings in the 1960s, and which applies to many physical situations, for instance in quantum optics and quantum information theory. Several of our results depend on the value of $p$: the structure of the model depends on the class of the prime $p$ modulo $4$ and $p=2$ requires special treatment. | [
{
"version": "v1",
"created": "Wed, 26 Jun 2024 15:09:59 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Crespo",
"Luis",
""
],
[
"Pelayo",
"Álvaro",
""
]
] |
2406.19274 | Orlando Luongo | Salvatore Capozziello, Alessio Lapponi, Orlando Luongo, Stefano Mancini | Preserving quantum information in $f(Q)$ cosmology | 16 pages, 3 figures, 2 tables | Eur.Phys.J.C 84 (2024) 10, 1081 | 10.1140/epjc/s10052-024-13449-3 | null | gr-qc astro-ph.CO quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The effects of cosmological expansion on quantum bosonic states are investigated, using quantum information theory. In particular, a generic Bogoliubov transformation of bosonic field modes is considered and the state change on a single mode is regarded as the effect of a quantum channel. Properties and capacities of this channel are thus explored in the framework of $f(Q)$ theories. As immediate result, we obtain that the information on a single-mode state appears better preserved, whenever the number of particles produced by the cosmological expansion is small. Hence, similarly to general relativity, we show that analogous particle productions result even if we consider symmetric teleparallel gravity theories. Thus, we investigate a power law $f(Q)$ model, leaving unaltered the effective gravitational coupling, and minimise the corresponding particle production. We thus show how to optimise the preservation of classical and quantum information, stored in a bosonic mode states in the remote past. Finally, we compare our findings with those obtained in general relativity. | [
{
"version": "v1",
"created": "Thu, 27 Jun 2024 15:43:09 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Capozziello",
"Salvatore",
""
],
[
"Lapponi",
"Alessio",
""
],
[
"Luongo",
"Orlando",
""
],
[
"Mancini",
"Stefano",
""
]
] |
2406.19303 | Tomasz Przezdziecki | Tomasz Przezdziecki, Jian-Rong Li | Compatibility of Drinfeld presentations for split affine Kac-Moody quantum symmetric pairs | 39 pages | null | null | null | math.QA math-ph math.MP math.RT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Let $(\mathbf{U}, \mathbf{U}^\imath)$ be a split affine quantum symmetric pair of type $\mathsf{B}_n^{(1)}, \mathsf{C}_n^{(1)}$ or $\mathsf{D}_n^{(1)}$. We prove factorization and coproduct formulae for the Drinfeld-Cartan operators $\Theta_i(z)$ in the Lu-Wang Drinfeld-type presentation, generalizing the type $\mathsf{A}_n^{(1)}$ result from [Prz23]. As an application, we show that a boundary analogue of the $q$-character map, defined via the spectra of these operators, is compatible with the usual $q$-character map. As an auxiliary result, we also produce explicit reduced expressions for the fundamental weights in the extended affine Weyl groups of classical types. | [
{
"version": "v1",
"created": "Thu, 27 Jun 2024 16:21:04 GMT"
},
{
"version": "v2",
"created": "Sat, 07 Jun 2025 22:12:13 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Przezdziecki",
"Tomasz",
""
],
[
"Li",
"Jian-Rong",
""
]
] |
2406.19532 | Ismail Alkhouri | Ismail Alkhouri, Cedric Le Denmat, Yingjie Li, Cunxi Yu, Jia Liu, Rongrong Wang, Alvaro Velasquez | Differentiable Quadratic Optimization For The Maximum Independent Set Problem | null | International Conference on Machine Learning 2025 | null | null | cs.DM cs.LG | http://creativecommons.org/licenses/by/4.0/ | Combinatorial Optimization (CO) addresses many important problems, including the challenging Maximum Independent Set (MIS) problem. Alongside exact and heuristic solvers, differentiable approaches have emerged, often using continuous relaxations of ReLU-based or quadratic objectives. Noting that an MIS in a graph is a Maximum Clique (MC) in its complement, we propose a new quadratic formulation for MIS by incorporating an MC term, improving convergence and exploration. We show that every maximal independent set corresponds to a local minimizer, derive conditions with respect to the MIS size, and characterize stationary points. To tackle the non-convexity of the objective, we propose optimizing several initializations in parallel using momentum-based gradient descent, complemented by an efficient MIS checking criterion derived from our theory. We dub our method as parallelized Clique-Informed Quadratic Optimization for MIS (pCQO-MIS). Our experimental results demonstrate the effectiveness of the proposed method compared to exact, heuristic, sampling, and data-centric approaches. Notably, our method avoids the out-of-distribution tuning and reliance on (un)labeled data required by data-centric methods, while achieving superior MIS sizes and competitive runtime relative to their inference time. Additionally, a key advantage of pCQO-MIS is that, unlike exact and heuristic solvers, the runtime scales only with the number of nodes in the graph, not the number of edges. Our code is available at the GitHub repository: https://github.com/ledenmat/pCQO-mis-benchmark/tree/refactor. | [
{
"version": "v1",
"created": "Thu, 27 Jun 2024 21:12:48 GMT"
},
{
"version": "v2",
"created": "Sun, 06 Oct 2024 21:31:27 GMT"
},
{
"version": "v3",
"created": "Wed, 16 Oct 2024 19:31:19 GMT"
},
{
"version": "v4",
"created": "Thu, 20 Feb 2025 21:02:23 GMT"
},
{
"version": "v5",
"created": "Thu, 01 May 2025 19:23:23 GMT"
},
{
"version": "v6",
"created": "Mon, 26 May 2025 18:09:18 GMT"
},
{
"version": "v7",
"created": "Fri, 06 Jun 2025 19:39:00 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Alkhouri",
"Ismail",
""
],
[
"Denmat",
"Cedric Le",
""
],
[
"Li",
"Yingjie",
""
],
[
"Yu",
"Cunxi",
""
],
[
"Liu",
"Jia",
""
],
[
"Wang",
"Rongrong",
""
],
[
"Velasquez",
"Alvaro",
""
]
] |
2406.19580 | Saeed Rashidi | Saeed Rashidi, William Won, Sudarshan Srinivasan, Puneet Gupta, and Tushar Krishna | FRED: Flexible REduction-Distribution Interconnect and Communication Implementation for Wafer-Scale Distributed Training of DNN Models | null | null | null | null | cs.AR cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Distributed Deep Neural Network (DNN) training is a technique to reduce the training overhead by distributing the training tasks into multiple accelerators, according to a parallelization strategy. However, high-performance compute and interconnects are needed for maximum speed-up and linear scaling of the system. Wafer-scale systems are a promising technology that allows for tightly integrating high-end accelerators with high-speed wafer-scale interconnects, making it an attractive platform for distributed training. However, the wafer-scale interconnect should offer high performance and flexibility for various parallelization strategies to enable maximum optimizations for compute and memory usage. In this paper, we propose FRED, a wafer-scale interconnect that is tailored for the high-BW requirements of wafer-scale networks and can efficiently execute communication patterns of different parallelization strategies. Furthermore, FRED supports in-switch collective communication execution that reduces the network traffic by approximately 2X. Our results show that FRED can improve the average end-to-end training time of ResNet-152, Transformer-17B, GPT-3, and Transformer-1T by 1.76X, 1.87X, 1.34X, and 1.4X, respectively when compared to a baseline waferscale 2D-Mesh fabric. | [
{
"version": "v1",
"created": "Fri, 28 Jun 2024 00:05:53 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 06:50:50 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Rashidi",
"Saeed",
""
],
[
"Won",
"William",
""
],
[
"Srinivasan",
"Sudarshan",
""
],
[
"Gupta",
"Puneet",
""
],
[
"Krishna",
"Tushar",
""
]
] |
2406.19705 | Kexiong Yu | Hang Zhao, Kexiong Yu, Yuhang Huang, Renjiao Yi, Chenyang Zhu, Kai Xu | DISCO: Efficient Diffusion Solver for Large-Scale Combinatorial Optimization Problems | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Combinatorial Optimization (CO) problems are fundamentally important in numerous real-world applications across diverse industries, characterized by entailing enormous solution space and demanding time-sensitive response. Despite recent advancements in neural solvers, their limited expressiveness struggles to capture the multi-modal nature of CO landscapes. While some research has shifted towards diffusion models, these models still sample solutions indiscriminately from the entire NP-complete solution space with time-consuming denoising processes, which limit their practicality for large problem scales. We propose DISCO, an efficient DIffusion Solver for large-scale Combinatorial Optimization problems that excels in both solution quality and inference speed. DISCO's efficacy is twofold: First, it enhances solution quality by constraining the sampling space to a more meaningful domain guided by solution residues, while preserving the multi-modal properties of the output distributions. Second, it accelerates the denoising process through an analytically solvable approach, enabling solution sampling with minimal reverse-time steps and significantly reducing inference time. DISCO delivers strong performance on large-scale Traveling Salesman Problems and challenging Maximal Independent Set benchmarks, with inference time up to 5.28 times faster than other diffusion alternatives. By incorporating a divide-and-conquer strategy, DISCO can well generalize to solve unseen-scale problem instances, even surpassing models specifically trained for those scales. | [
{
"version": "v1",
"created": "Fri, 28 Jun 2024 07:36:31 GMT"
},
{
"version": "v2",
"created": "Fri, 05 Jul 2024 02:44:25 GMT"
},
{
"version": "v3",
"created": "Fri, 19 Jul 2024 07:10:24 GMT"
},
{
"version": "v4",
"created": "Tue, 08 Oct 2024 15:59:13 GMT"
},
{
"version": "v5",
"created": "Mon, 21 Oct 2024 13:38:48 GMT"
},
{
"version": "v6",
"created": "Sun, 01 Jun 2025 03:35:31 GMT"
},
{
"version": "v7",
"created": "Mon, 09 Jun 2025 12:24:14 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Zhao",
"Hang",
""
],
[
"Yu",
"Kexiong",
""
],
[
"Huang",
"Yuhang",
""
],
[
"Yi",
"Renjiao",
""
],
[
"Zhu",
"Chenyang",
""
],
[
"Xu",
"Kai",
""
]
] |
2407.00976 | Nima Dehghani | Colleen J. Gillon, Cody Baker, Ryan Ly, Edoardo Balzani, Bingni W. Brunton, Manuel Schottdorf, Satrajit Ghosh, Nima Dehghani | ODIN: Open Data In Neurophysiology: Advancements, Solutions & Challenges | null | null | null | null | q-bio.NC | http://creativecommons.org/licenses/by-sa/4.0/ | Across the life sciences, an ongoing effort over the last 50 years has made data and methods more reproducible and transparent. This openness has led to transformative insights and vastly accelerated scientific progress. For example, structural biology and genomics have undertaken systematic collection and publication of protein sequences and structures over the past half-century, and these data have led to scientific breakthroughs that were unthinkable when data collection first began. We believe that neuroscience is poised to follow the same path, and that principles of open data and open science will transform our understanding of the nervous system in ways that are impossible to predict at the moment. To this end, new social structures along with active and open scientific communities are essential to facilitate and expand the still limited adoption of open science practices in our field. Unified by shared values of openness, we set out to organize a symposium for Open Data in Neuroscience (ODIN) to strengthen our community and facilitate transformative neuroscience research at large. In this report, we share what we learned during this first ODIN event. We also lay out plans for how to grow this movement, document emerging conversations, and propose a path toward a better and more transparent science of tomorrow. | [
{
"version": "v1",
"created": "Mon, 01 Jul 2024 05:26:30 GMT"
},
{
"version": "v2",
"created": "Sun, 08 Jun 2025 23:26:27 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Gillon",
"Colleen J.",
""
],
[
"Baker",
"Cody",
""
],
[
"Ly",
"Ryan",
""
],
[
"Balzani",
"Edoardo",
""
],
[
"Brunton",
"Bingni W.",
""
],
[
"Schottdorf",
"Manuel",
""
],
[
"Ghosh",
"Satrajit",
""
],
[
"Dehghani",
"Nima",
""
]
] |
2407.01101 | Alexander Natalchenko | Alexander Natalchenko, Arsenii Sagdeev | Packing Density of Sets With Only Two Nonmixed Gaps | 12 pages; a few modifications based on the reviews | null | null | null | math.CO math.NT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | For a finite set of integers such that the first few gaps between its consecutive elements equal $a$, while the remaining gaps equal $b$, we study dense packings of its translates on the line. We obtain an explicit lower bound on the corresponding optimal density, conjecture its tightness, and prove it in case one of the gap lengths, $a$ or $b$, appears only once. This is equivalent to a Motzkin problem on the independence ratio of certain integer distance graphs. | [
{
"version": "v1",
"created": "Mon, 01 Jul 2024 09:07:10 GMT"
},
{
"version": "v2",
"created": "Fri, 06 Jun 2025 20:57:05 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Natalchenko",
"Alexander",
""
],
[
"Sagdeev",
"Arsenii",
""
]
] |
2407.01887 | Fanzeng Xia | Fanzeng Xia, Hao Liu, Yisong Yue, Tongxin Li | Beyond Numeric Rewards: In-Context Dueling Bandits with LLM Agents | ACL 2025 Findings | null | null | null | cs.LG cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | In-Context Reinforcement Learning (ICRL) is a frontier paradigm to solve Reinforcement Learning (RL) problems in the foundation model era. While ICRL capabilities have been demonstrated in transformers through task-specific training, the potential of Large Language Models (LLMs) out-of-the-box remains largely unexplored. This paper investigates whether LLMs can generalize cross-domain to perform ICRL under the problem of Dueling Bandits (DB), a stateless preference-based RL setting. We find that the top-performing LLMs exhibit a notable zero-shot capacity for relative decision-making, which translates to low short-term weak regret across all DB environment instances by quickly including the best arm in duels. However, an optimality gap still exists between LLMs and classic DB algorithms in terms of strong regret. LLMs struggle to converge and consistently exploit even when explicitly prompted to do so, and are sensitive to prompt variations. To bridge this gap, we propose an agentic flow framework: LLM with Enhanced Algorithmic Dueling (LEAD), which integrates off-the-shelf DB algorithm support with LLM agents through fine-grained adaptive interplay. We show that LEAD has theoretical guarantees inherited from classic DB algorithms on both weak and strong regret. We validate its efficacy and robustness even with noisy and adversarial prompts. The design of such an agentic framework sheds light on how to enhance the trustworthiness of general-purpose LLMs generalized to in-context decision-making tasks. | [
{
"version": "v1",
"created": "Tue, 02 Jul 2024 02:18:14 GMT"
},
{
"version": "v2",
"created": "Tue, 15 Oct 2024 03:46:28 GMT"
},
{
"version": "v3",
"created": "Thu, 02 Jan 2025 13:49:59 GMT"
},
{
"version": "v4",
"created": "Mon, 09 Jun 2025 14:56:27 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Xia",
"Fanzeng",
""
],
[
"Liu",
"Hao",
""
],
[
"Yue",
"Yisong",
""
],
[
"Li",
"Tongxin",
""
]
] |
2407.02319 | Mengli Hu Dr. | Mengli Hu, Xingkai Cheng, Zhenqiao Huang, Junwei Liu | Catalogue of $C$-paired spin-momentum locking in antiferromagnetic systems | Main text: 11 pages, 4 figures | null | 10.1103/PhysRevX.15.021083 | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Antiferromagnetic materials (AFMs) have been gaining lots of attentions due to its great potential in spintronics devices and the recently discovered novel spin structure in the momentum space, i.e., $C$-paired spin-valley or spin-momentum locking (CSVL/CSML), where spins and valleys/momenta are locked to each other due to the crystal symmetry guaranteeing zero magnetization. Here, we systematically studied CSMLs and proposed a general theory and algorithm using little co-group and coset representatives, which reveals that 12 elementary kinds of CSMLs, determined by the geometric relation of spins and valleys and the essential symmetry guaranteeing zero magnetization, are sufficient to fully represent all possible CSMLs. By combining the proposed algorithm and high-throughput first-principles calculations, we predicted 38 magnetic point groups and identified 142 experimentally verified AFMs that can realize CSML. Besides predicting new materials, our theory can naturally reveal underlying mechanisms of CSMLs' responses to external fields. As an example, two qualitatively different types of piezomagnetism via occupation imbalance or spin tilting were predicted in RbV$_2$Te$_2$O. The algorithm and conclusions can be directly extended to the locking between valley/momentum and any other pseudo-vector degree of freedom, e.g., Berry curvature, as exemplified in RbV$_2$Te$_2$O and the new proposed piezo-Hall effect, where a strain can induce a non-zero anomalous Hall conductance. In addition, the proposed concept and methodology can be straightforwardly applied to other symmetry groups, such as spin group. | [
{
"version": "v1",
"created": "Tue, 02 Jul 2024 14:50:59 GMT"
},
{
"version": "v2",
"created": "Mon, 03 Mar 2025 12:22:46 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Hu",
"Mengli",
""
],
[
"Cheng",
"Xingkai",
""
],
[
"Huang",
"Zhenqiao",
""
],
[
"Liu",
"Junwei",
""
]
] |
2407.02549 | Mario Villaiz\'an-Vallelado | Mario Villaiz\'an-Vallelado and Matteo Salvatori and Carlos Segura and Ioannis Arapakis | Diffusion Models for Tabular Data Imputation and Synthetic Data Generation | 25 pages, 7 figures, 6 tables | null | 10.1145/3742435 | null | cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Data imputation and data generation have important applications for many domains, like healthcare and finance, where incomplete or missing data can hinder accurate analysis and decision-making. Diffusion models have emerged as powerful generative models capable of capturing complex data distributions across various data modalities such as image, audio, and time series data. Recently, they have been also adapted to generate tabular data. In this paper, we propose a diffusion model for tabular data that introduces three key enhancements: (1) a conditioning attention mechanism, (2) an encoder-decoder transformer as the denoising network, and (3) dynamic masking. The conditioning attention mechanism is designed to improve the model's ability to capture the relationship between the condition and synthetic data. The transformer layers help model interactions within the condition (encoder) or synthetic data (decoder), while dynamic masking enables our model to efficiently handle both missing data imputation and synthetic data generation tasks within a unified framework. We conduct a comprehensive evaluation by comparing the performance of diffusion models with transformer conditioning against state-of-the-art techniques, such as Variational Autoencoders, Generative Adversarial Networks and Diffusion Models, on benchmark datasets. Our evaluation focuses on the assessment of the generated samples with respect to three important criteria, namely: (1) Machine Learning efficiency, (2) statistical similarity, and (3) privacy risk mitigation. For the task of data imputation, we consider the efficiency of the generated samples across different levels of missing features. | [
{
"version": "v1",
"created": "Tue, 02 Jul 2024 15:27:06 GMT"
},
{
"version": "v2",
"created": "Mon, 02 Jun 2025 21:47:55 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Villaizán-Vallelado",
"Mario",
""
],
[
"Salvatori",
"Matteo",
""
],
[
"Segura",
"Carlos",
""
],
[
"Arapakis",
"Ioannis",
""
]
] |
2407.02725 | Yuya Mizuno | Yuya Mizuno, Dong Yang | Derived preprojective algebras and spherical twist functors | Reference to [BDL] and [HW] added | null | null | null | math.RT math.CT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study silting objects over derived preprojective algebras of acyclic quivers by giving a direct relationship between silting objects, spherical twist functors and mutations. Especially, for a Dynkin quiver, we establish a bijection between the elements of the braid group and the set of isomorphism classes of basic silting objects over the derived preprojective algebra. | [
{
"version": "v1",
"created": "Wed, 03 Jul 2024 00:35:18 GMT"
},
{
"version": "v2",
"created": "Fri, 20 Sep 2024 01:07:14 GMT"
},
{
"version": "v3",
"created": "Sat, 07 Jun 2025 08:25:04 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Mizuno",
"Yuya",
""
],
[
"Yang",
"Dong",
""
]
] |
2407.03127 | Udhav Fowdar | Udhav Fowdar, Henrique N. S\'a Earp | Flows of SU(2)-structures | 40 pages, to appear in Math. Z | null | null | null | math.DG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper initiates a classification programme of flows of $\mathrm{SU}(2)$-structures on $4$-manifolds which have short-time existence and uniqueness. Our approach adapts a representation-theoretic method originally due to Bryant in the context of $\mathrm{G}_2$ geometry. We show how this strategy can also be used to deduce the number of geometric flows of a given $H$-structure; we illustrate this in the $\mathrm{G}_2$, $\mathrm{Spin}(7)$ and $\mathrm{SU}(3)$ cases. Our investigation also leads us to derive explicit expressions for the Ricci and self-dual Weyl curvature in terms of the intrinsic torsion of the underlying $\mathrm{SU}(2)$-structure. We compute the first variation formulae of all the quadratic functionals in the torsion; these provide natural building blocks for $\mathrm{SU}(2)$ gradient flows. In particular, our results demonstrate that both the negative gradient flow of the Dirichlet energy of the intrinsic torsion and the Ricci harmonic flow are parabolic after a modified DeTurck's trick. | [
{
"version": "v1",
"created": "Wed, 03 Jul 2024 14:08:13 GMT"
},
{
"version": "v2",
"created": "Tue, 30 Jul 2024 15:13:34 GMT"
},
{
"version": "v3",
"created": "Sun, 08 Jun 2025 05:25:38 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Fowdar",
"Udhav",
""
],
[
"Earp",
"Henrique N. Sá",
""
]
] |
2407.03604 | Zhiyang Xu | Zhiyang Xu, Minqian Liu, Ying Shen, Joy Rimchala, Jiaxin Zhang, Qifan Wang, Yu Cheng, Lifu Huang | Modality-Specialized Synergizers for Interleaved Vision-Language Generalists | 8 Pages, interleaved instruction tuning, parameter-efficient tuning, image understanding, image generation | null | null | null | cs.CL cs.CV | http://creativecommons.org/licenses/by/4.0/ | Recent advancements in Vision-Language Models (VLMs) have led to the emergence of Vision-Language Generalists (VLGs) capable of understanding and generating both text and images. However, seamlessly generating an arbitrary sequence of text and images remains a challenging task for the current VLGs. One primary limitation lies in applying a unified architecture and the same set of parameters to simultaneously model discrete text tokens and continuous image features. Recent works attempt to tackle this fundamental problem by introducing modality-aware expert models. However, they employ identical architectures to process both text and images, disregarding the intrinsic inductive biases in these two modalities. In this work, we introduce MODALITY-SPECIALIZED SYNERGIZERS (MOSS), a novel design that efficiently optimizes existing unified architectures of VLGs with modality-specialized adaptation layers, i.e., a Convolutional LoRA for modeling the local priors of image patches and a Linear LoRA for processing sequential text. This design enables more effective modeling of modality-specific features while maintaining the strong cross-modal integration gained from pretraining. In addition, to improve the instruction-following capability on interleaved text-and-image generation, we introduce LEAFINSTRUCT, the first open-sourced interleaved instruction tuning dataset comprising 184,982 high-quality instances on more than 10 diverse domains. Extensive experiments show that VLGs integrated with M OSS achieve state-of-the-art performance, significantly surpassing baseline VLGs in complex interleaved generation tasks. Furthermore, our method exhibits strong generalizability on different VLGs. | [
{
"version": "v1",
"created": "Thu, 04 Jul 2024 03:28:22 GMT"
},
{
"version": "v2",
"created": "Sat, 07 Jun 2025 19:15:52 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Xu",
"Zhiyang",
""
],
[
"Liu",
"Minqian",
""
],
[
"Shen",
"Ying",
""
],
[
"Rimchala",
"Joy",
""
],
[
"Zhang",
"Jiaxin",
""
],
[
"Wang",
"Qifan",
""
],
[
"Cheng",
"Yu",
""
],
[
"Huang",
"Lifu",
""
]
] |
2407.03775 | Orlando Luongo | Alessio Belfiglio, S. Mahesh Chandran, Orlando Luongo, Stefano Mancini | Horizon entanglement area law from regular black hole thermodynamics | 14 pages, 7 figures | Phys.Rev.D 111 (2025) 2, 024013 | 10.1103/PhysRevD.111.024013 | null | gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the thermodynamics of regular black hole configurations via quantum analogs of entropy and energy -- namely, the entanglement entropy and entanglement energy -- near the event horizon of Bardeen and Hayward black holes. Following standard approaches, we introduce a quantum scalar field propagating in such black hole spacetimes and discretize the field degrees of freedom on a lattice of spherical shells. We observe that, at leading order, the entanglement entropy associated with the scalar field is proportional to Bekestein-Hawking entropy, while the corresponding entanglement energy scales proportionally to Komar energy. We then compute the heat capacity in both scenarios, discussing the black hole stability conditions and the possible appearance of second-order phase transitions. Finally, we extend our analysis to the black hole core, showing that in this sector entanglement energy serves as a valuable tool towards discriminating between singular and regular solutions. | [
{
"version": "v1",
"created": "Thu, 04 Jul 2024 09:34:41 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Belfiglio",
"Alessio",
""
],
[
"Chandran",
"S. Mahesh",
""
],
[
"Luongo",
"Orlando",
""
],
[
"Mancini",
"Stefano",
""
]
] |
2407.03883 | Xiaokun Luan | Xiaokun Luan, Xiyue Zhang, Jingyi Wang, Meng Sun | Protecting Deep Learning Model Copyrights with Adversarial Example-Free Reuse Detection | 12 pages, 5 figures | null | null | null | cs.CR cs.LG cs.SE | http://creativecommons.org/licenses/by/4.0/ | Model reuse techniques can reduce the resource requirements for training high-performance deep neural networks (DNNs) by leveraging existing models. However, unauthorized reuse and replication of DNNs can lead to copyright infringement and economic loss to the model owner. This underscores the need to analyze the reuse relation between DNNs and develop copyright protection techniques to safeguard intellectual property rights. Existing white-box testing-based approaches cannot address the common heterogeneous reuse case where the model architecture is changed, and DNN fingerprinting approaches heavily rely on generating adversarial examples with good transferability, which is known to be challenging in the black-box setting. To bridge the gap, we propose NFARD, a Neuron Functionality Analysis-based Reuse Detector, which only requires normal test samples to detect reuse relations by measuring the models' differences on a newly proposed model characterization, i.e., neuron functionality (NF). A set of NF-based distance metrics is designed to make NFARD applicable to both white-box and black-box settings. Moreover, we devise a linear transformation method to handle heterogeneous reuse cases by constructing the optimal projection matrix for dimension consistency, significantly extending the application scope of NFARD. To the best of our knowledge, this is the first adversarial example-free method that exploits neuron functionality for DNN copyright protection. As a side contribution, we constructed a reuse detection benchmark named Reuse Zoo that covers various practical reuse techniques and popular datasets. Extensive evaluations on this comprehensive benchmark show that NFARD achieves F1 scores of 0.984 and 1.0 for detecting reuse relationships in black-box and white-box settings, respectively, while generating test suites 2 ~ 99 times faster than previous methods. | [
{
"version": "v1",
"created": "Thu, 04 Jul 2024 12:21:59 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 03:31:53 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Luan",
"Xiaokun",
""
],
[
"Zhang",
"Xiyue",
""
],
[
"Wang",
"Jingyi",
""
],
[
"Sun",
"Meng",
""
]
] |
2407.04214 | Charles Wolock | Charles J. Wolock, Susan Jacob, Julia C. Bennett, Anna Elias-Warren, Jessica O'Hanlon, Avi Kenny, Nicholas P. Jewell, Andrea Rotnitzky, Stephen R. Cole, Ana A. Weil, Helen Y. Chu, Marco Carone | Investigating symptom duration using current status data: a case study of post-acute COVID-19 syndrome | The first two authors contributed equally to this work. Main text: 22 pages, 2 figure, 4 tables. Supplement: 23 pages, 14 figures, 0 tables. This update (v3) includes sensitivity analysis methodology and results | null | 10.1097/EDE.0000000000001882 | null | stat.AP stat.ME | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | For infectious diseases, characterizing symptom duration is of clinical and public health importance. Symptom duration may be assessed by surveying infected individuals and querying symptom status at the time of survey response. For example, in a SARS-CoV-2 testing program at the University of Washington, participants were surveyed at least $28$ days after testing positive and asked to report current symptom status. This study design yielded current status data: outcome measurements for each respondent consisted only of the time of survey response and a binary indicator of whether symptoms had resolved by that time. Such study design benefits from limited risk of recall bias, but analyzing the resulting data necessitates tailored statistical tools. Here, we review methods for current status data and describe a novel application of modern nonparametric techniques to this setting. The proposed approach is valid under weaker assumptions compared to existing methods, allows use of flexible machine learning tools, and handles potential survey nonresponse. From the university study, under an assumption that the survey response time is conditionally independent of symptom resolution time within strata of measured covariates, we estimate that 19% of participants experienced ongoing symptoms 30 days after testing positive, decreasing to 7% at 90 days. We assess the sensitivity of these results to deviations from conditional independence, finding the estimates to be more sensitive to assumption violations at 30 days compared to 90 days. Female sex, fatigue during acute infection, and higher viral load were associated with slower symptom resolution. | [
{
"version": "v1",
"created": "Fri, 05 Jul 2024 01:52:30 GMT"
},
{
"version": "v2",
"created": "Wed, 28 Aug 2024 18:01:44 GMT"
},
{
"version": "v3",
"created": "Mon, 17 Mar 2025 15:03:11 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Wolock",
"Charles J.",
""
],
[
"Jacob",
"Susan",
""
],
[
"Bennett",
"Julia C.",
""
],
[
"Elias-Warren",
"Anna",
""
],
[
"O'Hanlon",
"Jessica",
""
],
[
"Kenny",
"Avi",
""
],
[
"Jewell",
"Nicholas P.",
""
],
[
"Rotnitzky",
"Andrea",
""
],
[
"Cole",
"Stephen R.",
""
],
[
"Weil",
"Ana A.",
""
],
[
"Chu",
"Helen Y.",
""
],
[
"Carone",
"Marco",
""
]
] |
2407.04292 | Yiyang Huang | Yiyang Huang, Yuhui Hao, Bo Yu, Feng Yan, Yuxin Yang, Feng Min, Yinhe Han, Lin Ma, Shaoshan Liu, Qiang Liu, Yiming Gan | DaDu-Corki: Algorithm-Architecture Co-Design for Embodied AI-powered Robotic Manipulation | null | null | 10.1145/3695053.3731099 | null | cs.AR cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Embodied AI robots have the potential to fundamentally improve the way human beings live and manufacture. Continued progress in the burgeoning field of using large language models to control robots depends critically on an efficient computing substrate, and this trend is strongly evident in manipulation tasks. In particular, today's computing systems for embodied AI robots for manipulation tasks are designed purely based on the interest of algorithm developers, where robot actions are divided into a discrete frame basis. Such an execution pipeline creates high latency and energy consumption. This paper proposes \textsc{Corki}\xspace, an algorithm-architecture co-design framework for real-time embodied AI-powered robotic manipulation applications. We aim to decouple LLM inference, robotic control, and data communication in the embodied AI robots' compute pipeline. Instead of predicting action for one single frame, \textsc{Corki}\xspace predicts the trajectory for the near future to reduce the frequency of LLM inference. The algorithm is coupled with a hardware that accelerates transforming trajectory into actual torque signals used to control robots and an execution pipeline that parallels data communication with computation. \textsc{Corki}\xspace largely reduces LLM inference frequency by up to $5.1\times$, resulting in up to $5.9\times$ speed up. The success rate improvement can be up to 13.9\%. | [
{
"version": "v1",
"created": "Fri, 05 Jul 2024 06:54:26 GMT"
},
{
"version": "v2",
"created": "Thu, 01 Aug 2024 13:15:21 GMT"
},
{
"version": "v3",
"created": "Fri, 02 Aug 2024 01:59:22 GMT"
},
{
"version": "v4",
"created": "Sat, 16 Nov 2024 08:49:03 GMT"
},
{
"version": "v5",
"created": "Sun, 08 Jun 2025 15:01:21 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Huang",
"Yiyang",
""
],
[
"Hao",
"Yuhui",
""
],
[
"Yu",
"Bo",
""
],
[
"Yan",
"Feng",
""
],
[
"Yang",
"Yuxin",
""
],
[
"Min",
"Feng",
""
],
[
"Han",
"Yinhe",
""
],
[
"Ma",
"Lin",
""
],
[
"Liu",
"Shaoshan",
""
],
[
"Liu",
"Qiang",
""
],
[
"Gan",
"Yiming",
""
]
] |
2407.04874 | Kendall Mehling | Catie LeDesma, Kendall Mehling, and Murray Holland | Vector Atom Accelerometry in an Optical Lattice | null | Sci. Adv.11,eadt7480(2025) | 10.1126/sciadv.adt7480 | null | quant-ph cond-mat.quant-gas physics.atom-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We experimentally demonstrate two multidimensional atom interferometers capable of measuring both the magnitude and direction of applied inertial forces. These interferometers do not rely on the ubiquitous light-pulses of traditional atom sensors, but are instead built from an innovative design that operates entirely within the Bloch bands of an optical lattice formed by interfering laser beams. Through time-dependent control of the position of the lattice in three-dimensional space, we realize simultaneous Bloch oscillations in two dimensions, and a vector atomic Michelson interferometer. Fits to the observed Bloch oscillations demonstrate the measurement of an applied acceleration of $2g$ along two axes, where $g$ is the average gravitational acceleration at the Earth's surface. For the Michelson interferometer, we perform Bayesian inferencing from a 49-channel output by repeating experiments for selected examples of two-axis accelerations. We demonstrate the resulting accuracy and sensitivity for vector parameter estimation. Our acceleration can be measured from a single experimental run and does not require repeated shots to construct a fringe. We find the performance of our device to be near the quantum limit for the interferometer size and quantum detection efficiency of the atoms. We discuss the reconfigurability of the vector accelerometer and the pathway toward further sensitivity. | [
{
"version": "v1",
"created": "Fri, 05 Jul 2024 21:52:28 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"LeDesma",
"Catie",
""
],
[
"Mehling",
"Kendall",
""
],
[
"Holland",
"Murray",
""
]
] |
2407.04912 | Satoshi Usui | Takahiro Honma, Satoshi Usui | The stable category of Gorenstein-projective modules over a monomial algebra | v3: 44 pages, 2 figures, minor improvements to exposition, and corrected typos | null | null | null | math.RT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Let $\Lambda$ be an arbitrary monomial algebra. We investigate the stable category $\underline{\operatorname{Gproj}}^{\mathbb{Z}}\Lambda$ of graded Gorenstein-projective $\Lambda$-modules and the orbit category $\underline{\operatorname{Gproj}}^{\mathbb{Z}} \Lambda/(1)$ induced by $\underline{\operatorname{Gproj}}^{\mathbb{Z}}\Lambda$ and the degree shift functor $(1)$. We prove that $\underline{\operatorname{Gproj}}^{\mathbb{Z}}\Lambda$ is triangle equivalent to the bounded derived category of a path algebra of Dynkin type $\mathbb{A}$ and that $\underline{\operatorname{Gproj}}^{\mathbb{Z}}\Lambda/(1)$ is triangle equivalent to the stable module category of a self-injective Nakayama algebra. Both the path algebra and the self-injective Nakayama algebra will be given explicitly. The latter result provides an explicit description of the stable category of (ungraded) Gorenstein-projective $\Lambda$-modules. | [
{
"version": "v1",
"created": "Sat, 06 Jul 2024 01:36:25 GMT"
},
{
"version": "v2",
"created": "Fri, 13 Sep 2024 04:00:44 GMT"
},
{
"version": "v3",
"created": "Sat, 07 Jun 2025 03:36:50 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Honma",
"Takahiro",
""
],
[
"Usui",
"Satoshi",
""
]
] |
2407.05345 | Mikhail Shvartsman | Pavel B\v{e}l\'ik and Douglas P. Dokken and Mikhail M. Shvartsman | Review of Non-Equilibrium Thermodynamics And Statistical Mechanics of Vortex Gases in Tornado Theory | 20 pages | null | null | null | physics.flu-dyn physics.ao-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work puts into mathematical, statistical mechanical, and thermodynamical context the initial stages of the genesis of tornado-like vortices with the aim to be consistent with the current state of knowledge of the process of tornadogenesis. In particular, it discusses a mathematical foundation of the formation of coherent structures such as ``cusps'' and ``hairpins'' using variants of the nonlinear Schr\"odinger equation that arise via the Hasimoto transform of a vortex filament model. The behavior of such structures is then analyzed within a quasi-two-dimensional boundary layer model using the statistical mechanics of vortex gases to explain the rearrangement of cusps and other vertical vortex filaments into patches and possibly supercritical vortices. Non-equilibrium thermodynamics is used to obtain the entropic balance and the internal entropy production rate, and connect them to the turbulent heat flux. A formula for the non-equilibrium turbulent heat supply and formulas for the entropy supply and entropy production in the boundary layer are also provided. A relationship between the vorticity and the entropy gradient based on macroscopic fluctuations is given with implications to stretching and tilting of vorticity in the vertical direction. We conclude with some remarks on equivalence of Schr\"odinger and Gross-Pitaevskii equations in describing vortex filaments. | [
{
"version": "v1",
"created": "Sun, 07 Jul 2024 12:38:58 GMT"
},
{
"version": "v2",
"created": "Sat, 07 Jun 2025 15:48:05 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Bělík",
"Pavel",
""
],
[
"Dokken",
"Douglas P.",
""
],
[
"Shvartsman",
"Mikhail M.",
""
]
] |
2407.05958 | Gerhard Kirchmair | Aleksei Sharafiev, Mathieu Juan, Marco Cattaneo, Gerhard Kirchmair | Leveraging collective effects for thermometry in waveguide quantum electrodynamics | null | null | 10.1103/PhysRevLett.134.213602 | null | quant-ph cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | We report a proof-of-principle experiment for a new method of temperature measurements in waveguide quantum electrodynamics (wQED) experiments, allowing one to differentiate between global and local baths. The method takes advantage of collective states of two transmon qubits located in the center of a waveguide. The Hilbert space of such a system forms two separate subspaces (bright and dark) which are coupled differently to external noise sources. Measuring transmission through the waveguide allows one to extract separately the temperatures of the baths responsible for global and local excitations in the system. Such a system would allow for building a new type of primary temperature sensor capable of distinguishing between local and global baths. | [
{
"version": "v1",
"created": "Mon, 08 Jul 2024 13:56:48 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Sharafiev",
"Aleksei",
""
],
[
"Juan",
"Mathieu",
""
],
[
"Cattaneo",
"Marco",
""
],
[
"Kirchmair",
"Gerhard",
""
]
] |
2407.06279 | Flavio Del Santo | Flavio Del Santo, Gonzalo Manzano, and Caslav Brukner | Wigner's friend scenarios: on what to condition and how to verify the predictions | 8 pages + appendices | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Wigner's friend experiment and its modern extensions display the ambiguity of the quantum mechanical description regarding the assignment of quantum states. While the friend applies the state-update rule to the system upon observing an outcome of her measurement in a quantum system, Wigner describes the friend's measurement as a unitary evolution, resulting in an entangled state for the composite system of the friend and the system. In this respect, Wigner is often referred to as a "superobserver" who has the supreme technological ability to keep the friend's laboratory coherent. As such, it is often argued that he has the "correct" description of the state. Here we show that the situation is more symmetrical than is usually thought: there are different types of information that each of the observers has that the other fundamentally cannot have - they reside in different "bubbles" (in Calvalcanti's terminology). While this can explain why the objectivity of the state assignment is only relative to the bubble, we consider more elaborated situations in the form of a game in which the players can switch between bubbles. We find that, in certain circumstances, observers may be entitled to adopt and verify the state assignment from another bubble if they condition their predictions on \textit{all} information that is in principle available to them. | [
{
"version": "v1",
"created": "Mon, 08 Jul 2024 18:00:14 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 08:43:27 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Del Santo",
"Flavio",
""
],
[
"Manzano",
"Gonzalo",
""
],
[
"Brukner",
"Caslav",
""
]
] |
2407.06669 | Francisco Mart\'in-Rico | Francisco Mart\'in, Enrique Soriano-Salvador, Jos\'e Miguel Guerrero, Gorka Guardiola M\'uzquiz, Juan Carlos Manzanares, Francisco J. Rodr\'iguez | Towards a Robotic Intrusion Prevention System: Combining Security and Safety in Cognitive Social Robots | null | null | 10.1016/j.robot.2025.104959 | null | cs.RO | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Social Robots need to be safe and reliable to share their space with humans. This paper reports on the first results of a research project that aims to create more safe and reliable, intelligent autonomous robots by investigating the implications and interactions between cybersecurity and safety. We propose creating a robotic intrusion prevention system (RIPS) that follows a novel approach to detect and mitigate intrusions in cognitive social robot systems and other cyber-physical systems. The RIPS detects threats at the robotic communication level and enables mitigation of the cyber-physical threats by using System Modes to define what part of the robotic system reduces or limits its functionality while the system is compromised. We demonstrate the validity of our approach by applying it to a cognitive architecture running in a real social robot that preserves the privacy and safety of humans while facing several cyber attack situations. | [
{
"version": "v1",
"created": "Tue, 09 Jul 2024 08:44:27 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Martín",
"Francisco",
""
],
[
"Soriano-Salvador",
"Enrique",
""
],
[
"Guerrero",
"José Miguel",
""
],
[
"Múzquiz",
"Gorka Guardiola",
""
],
[
"Manzanares",
"Juan Carlos",
""
],
[
"Rodríguez",
"Francisco J.",
""
]
] |
2407.07969 | Francesco Benini | Francesco Benini, Victor Godet, Amartya Harsh Singh | Entanglement asymmetry in conformal field theory and holography | 29 pages plus appendices, 11 figures; v2: many improvements, figures redone, new app D, refs added; v3: some clarifications, published version | null | 10.1093/ptep/ptaf080 | SISSA 14/2024/FISI | hep-th cond-mat.stat-mech cond-mat.str-el quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Entanglement asymmetry is a measure of symmetry breaking in quantum subsystems, inspired by quantum information theory, particularly suited to study out-of-equilibrium states. We study the entanglement asymmetry of a class of excited "coherent states" in conformal quantum field theories with a U(1) symmetry, employing Euclidean path-integral methods with topological symmetry defects and the replica formalism. We compute, at leading order in perturbation theory, the asymmetry for a variety of subsystems, including finite spherical subregions in flat space, in finite volume, and at positive temperature. We also study its Lorentzian time evolution, showcasing the dynamical restoration of the symmetry due to thermalization, as well as the presence of a quantum Mpemba effect. Our results are universal, and apply in any number of dimensions. We also show that the perturbative entanglement asymmetry is related to the Fisher information metric, which has a known holographic dual called Hollands-Wald canonical energy, and that it is captured by the AdS bulk charge contained in the entanglement wedge. | [
{
"version": "v1",
"created": "Wed, 10 Jul 2024 18:08:27 GMT"
},
{
"version": "v2",
"created": "Fri, 04 Oct 2024 12:33:49 GMT"
},
{
"version": "v3",
"created": "Mon, 09 Jun 2025 07:52:57 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Benini",
"Francesco",
""
],
[
"Godet",
"Victor",
""
],
[
"Singh",
"Amartya Harsh",
""
]
] |
2407.08435 | Joachim Toft jto | P. K. Ratnakumar, Joachim Toft, Jasson Vindas | Non-isometric translation and modulation invariant Hilbert spaces | 13 pages. This is the third version of the document. We have mainly performed minor corrections compared to version 2. We observe the new title. The old title was "Non-isotropic translation and modulation invariant Hilbert spaces" | J. Math. Anal. App. 550 (2025), Article 129530 | 10.1016/j.jmaa.2025.129530 | null | math.FA | http://creativecommons.org/licenses/by/4.0/ | Let $\mathcal H$ be a Hilbert space of distributions on $\mathbf R^d$ which contains at least one non-zero element in $\mathscr D '(\mathbf R^d)$. If there is a constant $C_0>0$ such that $$ \nm {e^{i\scal \cdo \xi}f(\cdo -x)}{\mathcal H}\le C_0\nm f{\mathcal H}, \qquad f\in \mathcal H ,\ x,\xi \in \mathbf R^d, $$ then we prove that $\maclH = L^2(\mathbf R^d)$, with equivalent norms. | [
{
"version": "v1",
"created": "Thu, 11 Jul 2024 12:19:49 GMT"
},
{
"version": "v2",
"created": "Mon, 04 Nov 2024 18:15:54 GMT"
},
{
"version": "v3",
"created": "Fri, 07 Feb 2025 08:02:33 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Ratnakumar",
"P. K.",
""
],
[
"Toft",
"Joachim",
""
],
[
"Vindas",
"Jasson",
""
]
] |
2407.09056 | Yongfeng Zhu | Yongfeng Zhu, Weifeng Zhuang, Chen Qian, Yunheng Ma, Dong E. Liu, Manqi Ruan, Chen Zhou | A Novel Quantum Realization of Jet Clustering in High-Energy Physics Experiments | null | null | 10.1016/j.scib.2024.12.020 | null | quant-ph hep-ex | http://creativecommons.org/publicdomain/zero/1.0/ | Exploring the application of quantum technologies to fundamental sciences holds the key to fostering innovation for both sides. In high-energy particle collisions, quarks and gluons are produced and immediately form collimated particle sprays known as jets. Accurate jet clustering is crucial as it retains the information of the originating quark or gluon and forms the basis for studying properties of the Higgs boson, which underlies teh mechanism of mass generation for subatomic particles. For the first time, by mapping collision events into graphs--with particles as nodes and their angular separations as edges--we realize jet clustering using the Quantum Approximate Optimization Algorithm (QAOA), a hybrid quantum-classical algorithm for addressing classical combinatorial optimization problems with available quantum resources. Our results, derived from 30 qubits on quantum computer simulator and 6 qubits on quantum computer hardware, demonstrate that jet clustering performance with QAOA is comparable with or even better than classical algorithms for a small-sized problem. This study highlights the feasibility of quantum computing to revolutionize jet clustering, bringing the practical application of quantum computing in high-energy physics experiments one step closer. | [
{
"version": "v1",
"created": "Fri, 12 Jul 2024 07:26:22 GMT"
},
{
"version": "v2",
"created": "Thu, 03 Oct 2024 01:53:27 GMT"
},
{
"version": "v3",
"created": "Sat, 07 Jun 2025 09:59:00 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Zhu",
"Yongfeng",
""
],
[
"Zhuang",
"Weifeng",
""
],
[
"Qian",
"Chen",
""
],
[
"Ma",
"Yunheng",
""
],
[
"Liu",
"Dong E.",
""
],
[
"Ruan",
"Manqi",
""
],
[
"Zhou",
"Chen",
""
]
] |
2407.09188 | Francesco Macheda | Francesco Macheda and Paolo Barone and Francesco Mauri | First principles calculations of dynamical Born effective charges, quadrupoles and higher order terms from the charge response in large semiconducting and metallic systems | Reuploaded with CC BY licence | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Within the context of first principles techniques we present a theoretical and computational framework to quickly determine, at finite momentum, the self-consistent (longitudinal) charge response to an external perturbation, that enters the determination of the scattering cross section of inelastic scattering processes such as EELS. We also determine the (tranverse) charge response computed in short-circuit condition. The all-order quasimomentum expansion of the tranverse charge response to an atomic displacement are the Born effective charges, quadrupoles, octupoles etc. We demonstrate that the transverse charge response can be related to the longitudinal one via a well-defined long-range dielectric function. Our advancements lead to an efficient use of perturbation theory. Due to its more favorable scaling, our method provides an interesting computational alternative to the use of the 2n+1 theorem, especially for semiconductors and metals with large unit cells. For semiconductors, we compute the piezoelectric properties of a large cell solid-solution of semiconducting hafniun oxide containing 96 atoms. We here show that the clamped ion piezoelectric response can be decomposed into real-space localized contributions that mostly depend on the chemical environment, paving the way for the use of machine-learning techniques in the material search for optimized piezoelectrics. We further apply our methodology to determine the density response of metals. Here, the leading terms of the charge expansion are related to the Fermi energy shift of the potential and by Born effective charges which do not sum to zero over the atoms. We apply our developments to the TEM-EELS spectroscopy of lithium intercalated graphites, where we find that the use of the atomic form-factor in the long-wavelength limit does not take into account for the anisotropy of the atomic chemical bonding. | [
{
"version": "v1",
"created": "Fri, 12 Jul 2024 11:40:48 GMT"
},
{
"version": "v2",
"created": "Sat, 07 Jun 2025 14:06:28 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Macheda",
"Francesco",
""
],
[
"Barone",
"Paolo",
""
],
[
"Mauri",
"Francesco",
""
]
] |
2407.09207 | Andrea Dubla Dr | Pengzhong Lu, Rafet Kavak, Andrea Dubla, Silvia Masciocchi and Ilya Selyuzhenkov | Quantification of the low-$p_{\rm T}$ pion excess in heavy-ion collisions at the LHC and top RHIC energy | 9 pages, 6 figures | NUCL SCI TECH 36, 142 (2025) | 10.1007/s41365-025-01718-z | null | hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While the abundances of the final state hadrons in relativistic heavy-ion collisions are rather well described by the thermal particle production, the shape of the transverse momentum, $p_{\rm T}$, distribution below $p_{\rm T}\approx500$ MeV$/c$, is still poorly understood. We propose a procedure to quantify the model-to-data differences using Bayesian inference techniques, which allows for consistent treatment of the experimental uncertainties and tests the completeness of the available hydrodynamic frameworks. Using relativistic fluid framework Fluid${\it u}$M with PCE coupled to \trento initial state and FastReso decays, we analyse \pt distribution of identified charged hadrons measured in heavy-ion collisions at top RHIC and the LHC energies, and identify an excess of pions produced below $p_{\rm T}\approx500$ MeV$/c$. Our results provide new input for the interpretation of the pion excess as either missing components in the thermal particle yield description or as an evidence for a different particle production mechanism. | [
{
"version": "v1",
"created": "Fri, 12 Jul 2024 12:14:24 GMT"
},
{
"version": "v2",
"created": "Fri, 06 Jun 2025 07:31:19 GMT"
},
{
"version": "v3",
"created": "Mon, 09 Jun 2025 07:12:02 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Lu",
"Pengzhong",
""
],
[
"Kavak",
"Rafet",
""
],
[
"Dubla",
"Andrea",
""
],
[
"Masciocchi",
"Silvia",
""
],
[
"Selyuzhenkov",
"Ilya",
""
]
] |
2407.09299 | Fangyuan Mao | Fangyuan Mao, Jilin Mei, Shun Lu, Fuyang Liu, Liang Chen, Fangzhou Zhao, Yu Hu | PID: Physics-Informed Diffusion Model for Infrared Image Generation | Accepted by Pattern Recognition | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Infrared imaging technology has gained significant attention for its reliable sensing ability in low visibility conditions, prompting many studies to convert the abundant RGB images to infrared images. However, most existing image translation methods treat infrared images as a stylistic variation, neglecting the underlying physical laws, which limits their practical application. To address these issues, we propose a Physics-Informed Diffusion (PID) model for translating RGB images to infrared images that adhere to physical laws. Our method leverages the iterative optimization of the diffusion model and incorporates strong physical constraints based on prior knowledge of infrared laws during training. This approach enhances the similarity between translated infrared images and the real infrared domain without increasing extra training parameters. Experimental results demonstrate that PID significantly outperforms existing state-of-the-art methods. Our code is available at https://github.com/fangyuanmao/PID. | [
{
"version": "v1",
"created": "Fri, 12 Jul 2024 14:32:30 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 14:35:16 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Mao",
"Fangyuan",
""
],
[
"Mei",
"Jilin",
""
],
[
"Lu",
"Shun",
""
],
[
"Liu",
"Fuyang",
""
],
[
"Chen",
"Liang",
""
],
[
"Zhao",
"Fangzhou",
""
],
[
"Hu",
"Yu",
""
]
] |
2407.09590 | Zeliang Zhang | Zeliang Zhang, Xiaodong Liu, Hao Cheng, Chenliang Xu, Jianfeng Gao | Diversifying the Expert Knowledge for Task-Agnostic Pruning in Sparse Mixture-of-Experts | Findings of ACL 25. Fix typos. Correct and add more details on the expert similarity measurement | null | null | null | cs.CL cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | By increasing model parameters but activating them sparsely when performing a task, the use of Mixture-of-Experts (MoE) architecture significantly improves the performance of Large Language Models (LLMs) without increasing the inference cost. However, the memory consumption due to the growing number of experts presents a challenge to the deployment of these models in many real world settings. Our empirical study reveals that some experts encode redundant knowledge during pre-training. We thus propose a method of grouping and pruning similar experts to improve the model's parameter efficiency. We validate the effectiveness of our method by pruning three state-of-the-art MoE architectures, including Mixtral, Deepseek-MoE, and Qwen. The evaluation shows that our method outperforms other model pruning methods on a range of natural language tasks. We will release our code to facilitate future research. | [
{
"version": "v1",
"created": "Fri, 12 Jul 2024 17:25:02 GMT"
},
{
"version": "v2",
"created": "Tue, 17 Sep 2024 13:48:50 GMT"
},
{
"version": "v3",
"created": "Sat, 19 Oct 2024 21:46:33 GMT"
},
{
"version": "v4",
"created": "Sun, 08 Jun 2025 17:00:34 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Zhang",
"Zeliang",
""
],
[
"Liu",
"Xiaodong",
""
],
[
"Cheng",
"Hao",
""
],
[
"Xu",
"Chenliang",
""
],
[
"Gao",
"Jianfeng",
""
]
] |
2407.09881 | Yuta Nozaki | Michel Boileau, Teruaki Kitano, Yuta Nozaki | On the genera of symmetric unions of knots | 28 pages, 8 figures | null | null | null | math.GT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the study of ribbon knots, Lamm introduced symmetric unions inspired by earlier work of Kinoshita and Terasaka. We show an identity between the twisted Alexander polynomials of a symmetric union and its partial knot. As a corollary, we obtain an inequality concerning their genera. It is known that there exists an epimorphism between their knot groups, and thus our inequality provides a positive answer to an old problem of Jonathan Simon in this case. Our formula also offers a useful condition to constrain possible symmetric union presentations of a given ribbon knot. It is an open question whether every ribbon knot is a symmetric union. | [
{
"version": "v1",
"created": "Sat, 13 Jul 2024 13:10:40 GMT"
},
{
"version": "v2",
"created": "Sun, 08 Jun 2025 07:07:17 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Boileau",
"Michel",
""
],
[
"Kitano",
"Teruaki",
""
],
[
"Nozaki",
"Yuta",
""
]
] |
2407.10089 | Xinyi Fang | Xinyi Fang and Mengyang Gu | The inverse Kalman filter | 17 pages, 8 figures, 2 tables | null | null | null | stat.ME stat.AP stat.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce the inverse Kalman filter, which enables exact matrix-vector multiplication between a covariance matrix from a dynamic linear model and any real-valued vector with linear computational cost. We integrate the inverse Kalman filter with the conjugate gradient algorithm, which substantially accelerates the computation of matrix inversion for a general form of covariance matrix, where other approximation approaches may not be directly applicable. We demonstrate the scalability and efficiency of the proposed approach through applications in nonparametric estimation of particle interaction functions, using both simulations and cell trajectories from microscopy data. | [
{
"version": "v1",
"created": "Sun, 14 Jul 2024 05:30:22 GMT"
},
{
"version": "v2",
"created": "Sat, 03 Aug 2024 21:49:31 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Feb 2025 23:29:02 GMT"
},
{
"version": "v4",
"created": "Sat, 07 Jun 2025 19:22:33 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Fang",
"Xinyi",
""
],
[
"Gu",
"Mengyang",
""
]
] |
2407.11239 | Ajay Jaiswal | Ajay Jaiswal, Yifan Wang, Lu Yin, Shiwei Liu, Runjin Chen, Jiawei Zhao, Ananth Grama, Yuandong Tian, Zhangyang Wang | From Low Rank Gradient Subspace Stabilization to Low-Rank Weights: Observations, Theories, and Applications | Accepted in ICML 2025 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Large Language Models' (LLMs) weight matrices can often be expressed in low-rank form with potential to relax memory and compute resource requirements. Unlike prior efforts that focus on developing novel matrix decompositions, in this work we study the non-uniform low-rank properties of weight matrices in LLMs through the lens of stabilizing gradient subspace. First, we provide a theoretical framework to understand the stabilization of gradient subspaces through Hessian analysis. Second, we empirically establish an important relationship between gradient dynamics and low-rank expressiveness of weight matrices. Our findings reveal that different LLM components exhibit varying levels of converged low-rank structures, necessitating variable rank reduction across them to minimize drop in performance due to compression. Drawing on this result, we present Weight Low-Rank Projection(WeLore) that unifies weight compression and memory-efficient fine-tuning into one, in a data-agnostic and one-shot manner. When used as a compression technique, WeLore categorizes weight matrices into Low-rank Components (LRCs) and Non-Low-rank Components (N-LRCs) and suitably encodes them for minimum performance loss. Our gradient dynamics perspective illustrates that LRCs tend to have better fine-tuning capabilities and their standalone fine-tuning can closely mimic and sometimes outperform the training loss trajectory and performance of full fine-tuning with notable memory and compute footprint reduction. Codes are available at https://github.com/VITA-Group/WeLore. | [
{
"version": "v1",
"created": "Mon, 15 Jul 2024 21:05:20 GMT"
},
{
"version": "v2",
"created": "Sat, 07 Jun 2025 06:59:36 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Jaiswal",
"Ajay",
""
],
[
"Wang",
"Yifan",
""
],
[
"Yin",
"Lu",
""
],
[
"Liu",
"Shiwei",
""
],
[
"Chen",
"Runjin",
""
],
[
"Zhao",
"Jiawei",
""
],
[
"Grama",
"Ananth",
""
],
[
"Tian",
"Yuandong",
""
],
[
"Wang",
"Zhangyang",
""
]
] |
2407.12124 | Xin Li | Xin Li, Robert J. Huebner, Margot L.K. Williams, Jessica Sawyer, Mark Peifer, John B. Wallingford, and D. Thirumalai | Emergence of cellular nematic order is a conserved feature of gastrulation in animal embryos | Main text: 39 pages, 10 figures. SI: 14 figures and 1 table | null | null | null | cond-mat.soft physics.bio-ph q-bio.TO | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Cells undergo dramatic morphological changes during embryogenesis, yet how these changes affect the formation of ordered tissues remains elusive. Here, we show that a phase transition leading to the formation of a nematic liquid crystal state during gastrulation in the development of embryos of fish, frogs, and fruit flies occurs by a common mechanism despite substantial differences between these evolutionarily distant animals. Importantly, nematic order forms early before any discernible changes in the shapes of cells. All three species exhibit similar propagation of the nematic phase, reminiscent of nucleation and growth mechanisms. The spatial correlations in the nematic phase in the notochord region are long-ranged and follow a similar power-law decay (y~$x^{-\alpha}$ ) with $\alpha$ less than unity, indicating a common underlying physical mechanism. To explain the common physical mechanism, we created a theoretical model that not only explains the experimental observations but also predicts that the nematic phase should be disrupted upon loss of planar cell polarity (frog), cell adhesion (frog), and notochord boundary formation (zebrafish). Gene knockdown or mutational studies confirm the theoretical predictions. The combination of experiments and theory provides a unified framework for understanding the potentially universal features of metazoan embryogenesis, in the process shedding light on the advent of ordered structures during animal development. | [
{
"version": "v1",
"created": "Tue, 16 Jul 2024 19:20:01 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 03:43:58 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Li",
"Xin",
""
],
[
"Huebner",
"Robert J.",
""
],
[
"Williams",
"Margot L. K.",
""
],
[
"Sawyer",
"Jessica",
""
],
[
"Peifer",
"Mark",
""
],
[
"Wallingford",
"John B.",
""
],
[
"Thirumalai",
"D.",
""
]
] |
2407.12429 | Andrey Kudlis | R.V. Cherbunin, A. Liubomirov, D. Novokreschenov, A. Kudlis, A.V. Kavokin | Quantum beats of a macroscopic polariton condensate in real space | null | null | null | null | cond-mat.mes-hall physics.optics | http://creativecommons.org/licenses/by/4.0/ | We experimentally observe harmonic oscillations in a bosonic condensate of exciton-polaritons confined within an elliptical trap. These oscillations arise from quantum beats between two size-quantized states of the condensate, split in energy due to the trap's ellipticity. By precisely targeting specific spots inside the trap with nonresonant laser pulses, we control frequency, amplitude, and phase of these quantum beats. The condensate wave function dynamics is visualized on a streak camera and mapped to the Bloch sphere, demonstrating Hadamard and Pauli-Z operations. We conclude that a qubit based on a superposition of these two polariton states would exhibit a coherence time exceeding the lifetime of an individual exciton-polariton by at least two orders of magnitude. | [
{
"version": "v1",
"created": "Wed, 17 Jul 2024 09:33:53 GMT"
},
{
"version": "v2",
"created": "Sat, 24 Aug 2024 15:47:51 GMT"
},
{
"version": "v3",
"created": "Mon, 09 Jun 2025 10:06:18 GMT"
}
] | 2025-06-10T00:00:00 | [
[
"Cherbunin",
"R. V.",
""
],
[
"Liubomirov",
"A.",
""
],
[
"Novokreschenov",
"D.",
""
],
[
"Kudlis",
"A.",
""
],
[
"Kavokin",
"A. V.",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.