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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2506.04659 | Mukesh Rajmohan | Mukesh Rajmohan, Smit Desai and Sanchari Das | Multi-Tool Analysis of User Interface & Accessibility in Deployed Web-Based Chatbots | 9 pages, 6 figures. Submitted to ACM Conversational User Interfaces (CUI) 2025 | null | null | null | cs.HC | http://creativecommons.org/licenses/by/4.0/ | In this work, we present a multi-tool evaluation of 106 deployed web-based chatbots, across domains like healthcare, education and customer service, comprising both standalone applications and embedded widgets using automated tools (Google Lighthouse, PageSpeed Insights, SiteImprove Accessibility Checker) and manual audits (Microsoft Accessibility Insights). Our analysis reveals that over 80% of chatbots exhibit at least one critical accessibility issue, and 45% suffer from missing semantic structures or ARIA role misuse. Furthermore, we found that accessibility scores correlate strongly across tools (e.g., Lighthouse vs PageSpeed Insights, r = 0.861), but performance scores do not (r = 0.436), underscoring the value of a multi-tool approach. We offer a replicable evaluation insights and actionable recommendations to support the development of user-friendly conversational interfaces. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:02:59 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Rajmohan",
"Mukesh",
""
],
[
"Desai",
"Smit",
""
],
[
"Das",
"Sanchari",
""
]
] |
2506.04660 | Yi Xu | Yi Xu, Farzin Lotfi-Jam and Mustafa Faruki | Adaptive recycled plastic architecture: Vacuum-Sealed Chainmail Structures Through Computational Design | Accepted manuscript. Published in International Journal of Architectural Computing, April 2025 | null | 10.1177/14780771251335110 | null | cs.CE cond-mat.mtrl-sci | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The construction industry is a major consumer of raw materials, accounting for nearly half of global material usage annually, while generating significant waste that poses sustainability challenges. This paper explores the untapped potential of recycled plastics as a primary construction material, leveraging their lightweight, flexible, and customizable properties for advanced applications in modular chainmail systems. Through a computational workflow, the study optimizes the design, testing, and fabrication of vacuum-sealed chainmail structures composed of recycled plastic filaments, demonstrating their adaptability and structural performance for architectural use.
Key contributions include a novel methodology for integrating recycled plastic filaments into chainmail geometries, validated through 2D sectional testing, 3D shell structure generation, and physical modeling under vacuum constraints. The research identifies the rectangular chainmail configuration as the most efficient and adaptable, achieving superior deformation capacity, material efficiency, and load-bearing performance. Optimization strategies for temporary structures highlight practical deployment potential, balancing material savings, usable area, and water drainage efficiency.
The findings offer a foundation for innovative applications in extreme conditions, including disaster-prone areas, high-altitude environments, underwater platforms, and extraterrestrial habitats. These applications leverage the lightweight, adaptable, and durable properties of recycled plastics and modular chainmail systems, bridging the gap between waste management and high-performance design while addressing unique challenges in harsh and resource-constrained environments. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:03:32 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Xu",
"Yi",
""
],
[
"Lotfi-Jam",
"Farzin",
""
],
[
"Faruki",
"Mustafa",
""
]
] |
2506.04661 | Haitian Shang | Haitian Shang, Wei Zhao, Xiaoyu Hong, and Xu-zhi Hu | Analysis of Jet Dynamics and Collimation Characteristics of 0241+622 on Parsec Scales | 18 pages, 11 figures, accepted for publication in The Astrophysical Journal | null | 10.3847/1538-4357/addd10 | null | astro-ph.HE astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We conducted a detailed analysis of the jet structure and dynamics of the source 0241+622 on milliarcsecond (mas) scales. We stacked images from multiple epochs to better recover the crosssection of the jet. By analyzing the relationship between jet width and distance, we observed that the jet exhibits a parabolic shape from the core, spanning a region from 0.12 to 6.1 mas. This structure suggests the acceleration and collimation processes of the jet. Beyond 18 mas from the core, the jet adopts a conical shape, and the expansion speed of the jet becomes faster within the range from 4500 to 6500 mas. We obtained the core shift of this source using five pairs of data from VLBA at 1.6 GHz to 43 GHz. Based on previous studies, through proper motion analysis of the jet components, we estimated the angle between the jet and the line of sight to be approximately 65.7{\deg}, so 1 mas corresponds to 0.95 pc (de-projected distance). We then obtained the velocity field of the source within 3.14 mas from the central black hole and found that the jet exhibits accelerated motion within this range. At approximately 6.1 mas from the core, we observed that the jet width begins to decrease, which we identified as possibly corresponding to the Bondi radius of this source. The reduction in jet width may be related to changes in the external environmental pressure, particularly within the Bondi radius, indicating that the jet dynamics and collimation characteristics are strongly influenced by the surrounding medium conditions. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:03:50 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Shang",
"Haitian",
""
],
[
"Zhao",
"Wei",
""
],
[
"Hong",
"Xiaoyu",
""
],
[
"Hu",
"Xu-zhi",
""
]
] |
2506.04662 | Ewelina Nawara | Ewelina Nawara | The Hesse pencil of plane curves and osculating conics | 10 pages | null | null | null | math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we revisit the classical problem of determining osculating conics and sextactic points for a given algebraic curve. Our focus is on a particular family of plane cubic curves known as the Hesse pencil. By employing classical tools from projective differential geometry, we derive explicit coordinates for these special points. The resulting formulas not only clarify previous approaches but also lead to the construction of new families of free and nearly free curves, extending recent findings the freeness of curves. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:05:12 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Nawara",
"Ewelina",
""
]
] |
2506.04663 | Takumi Kobori | Takumi Kobori, Taichi Kosugi, Hirofumi Nishi, Synge Todo and Yu-ichiro Matsushita | Accelerated spin-adapted ground state preparation with non-variational quantum algorithms | 12 pages, 8 figures | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Various methods have been explored to prepare the spin-adapted ground state, the lowest energy state within the Hilbert space constrained by externally specified values of the total spin magnitude and the spin-$z$ component. In such problem settings, variational and non-variational methods commonly incorporate penalty terms into the original Hamiltonian to enforce the desired constraints. While in variational approaches, only $O(n_{\textrm{spin}}^2)$ measurements are required for the calculation of the penalty terms for the total spin magnitude, non-variational approaches, such as probabilistic imaginary-time evolution or adiabatic time evolution, are expected to be more computationally intensive, requiring $O(n_{\textrm{spin}}^4)$ gates naively. This paper proposes a new procedure based on non-variational quantum algorithms to obtain the spin-adapted ground state. The proposed method consists of two steps: the first step is to prepare a spin-magnitude adapted state and the second step is post-processing for the desired $S_z$. By separating into two steps, the procedure achieves the desired spin-adapted ground state while reducing the number of penalty terms from $O(n_{\textrm{spin}}^4)$ to $O(n_{\textrm{spin}}^2)$. We conducted numerical experiments for spin-1/2 Heisenberg ring models and manganese trimer systems. The results confirmed the effectiveness of our method, demonstrating a significant reduction in gate complexity and validating its practical usefulness. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:16:17 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Kobori",
"Takumi",
""
],
[
"Kosugi",
"Taichi",
""
],
[
"Nishi",
"Hirofumi",
""
],
[
"Todo",
"Synge",
""
],
[
"Matsushita",
"Yu-ichiro",
""
]
] |
2506.04664 | Bimal Kumar Ray | Bimal Kumar Ray | A Fast Unsupervised Scheme for Polygonal Approximation | null | null | null | null | cs.GR cs.CG cs.CV | http://creativecommons.org/licenses/by/4.0/ | This paper proposes a fast and unsupervised scheme for a polygonal approximation of a closed digital curve. It is demonstrated that the approximation scheme is faster than state-of-the-art approximation and is competitive with the same in Rosin's measure and in its aesthetic aspect. The scheme comprises of three phases: initial segmentation, iterative vertex insertion, and iterative merging, followed by vertex adjustment. The initial segmentation is used to detect sharp turnings - the vertices that seemingly have high curvature. It is likely that some of important vertices with low curvature might have been missed out at the first phase and so iterative vertex insertion is used to add vertices in a region where the curvature changes slowly but steadily. The initial phase may pick up some undesirable vertices and so merging is used to eliminate the redundant vertices. Finally, vertex adjustment is used to facilitate enhancement in the aesthetic look of the approximation. The quality of the approximations is measured using Rosin's measure. The robustness of the proposed scheme with respect to geometric transformation is observed. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:18:48 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Ray",
"Bimal Kumar",
""
]
] |
2506.04665 | Rian Neogi | Rian Neogi, Kanstantsin Pashkovich, Chaitanya Swamy | An O(log log n)-approximate budget feasible mechanism for subadditive valuations | null | null | null | null | cs.GT | http://creativecommons.org/licenses/by/4.0/ | In budget-feasible mechanism design, there is a set of items $U$, each owned by a distinct seller. The seller of item $e$ incurs a private cost $\overline{c}_e$ for supplying her item. A buyer wishes to procure a set of items from the sellers of maximum value, where the value of a set $S\subseteq U$ of items is given by a valuation function $v:2^U\to \mathbb{R}_+$. The buyer has a budget of $B \in \mathbb{R}_+$ for the total payments made to the sellers. We wish to design a mechanism that is truthful, that is, sellers are incentivized to report their true costs, budget-feasible, that is, the sum of the payments made to the sellers is at most the budget $B$, and that outputs a set whose value is large compared to $\text{OPT}:=\max\{v(S):\overline{c}(S)\le B,S\subseteq U\}$.
Budget-feasible mechanism design has been extensively studied, with the literature focussing on (classes of) subadditive valuation functions, and various polytime, budget-feasible mechanisms, achieving constant-factor approximation, have been devised for the special cases of additive, submodular, and XOS valuations. However, for general subadditive valuations, the best-known approximation factor achievable by a polytime budget-feasible mechanism (given access to demand oracles) was only $O(\log n / \log \log n)$, where $n$ is the number of items.
We improve this state-of-the-art significantly by designing a budget-feasible mechanism for subadditive valuations that \emph{achieves a substantially-improved approximation factor of $O(\log\log n)$ and runs in polynomial time, given access to demand oracles.} | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:21:33 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Neogi",
"Rian",
""
],
[
"Pashkovich",
"Kanstantsin",
""
],
[
"Swamy",
"Chaitanya",
""
]
] |
2506.04666 | Evangelos Almpanis | Evangelos Almpanis | Axion-mediated photon-to-photon transitions in high finesse dielectric resonators | null | null | null | null | hep-ph cond-mat.mtrl-sci hep-th | http://creativecommons.org/licenses/by/4.0/ | Axions are hypothetical particles that could address both the strong charge-parity problem in quantum chromodynamics and the enigmatic nature of dark matter. However, if axions exist, their mass remains unknown, and they are expected to interact very weakly with the electromagnetic field, which explains why they have not been detected yet. This study proposes a way to substantially augment the axion-photon interaction by confining the photons within high-quality-factor dielectric resonators, increasing their intensity and lifetime and thus the possibility of interacting with axions in the background. In view of this, we study resonant axion-mediated photonic transitions in millimeter-sized spherical dielectric resonators, based on fully analytical calculations to the first order in perturbation theory. Such resonators exhibit high lifetime Mie resonances in the microwave part of the spectrum, with a separation that can be tailored with the radius of the sphere to match the expected axion frequency, allowing axion-mediated photonic transitions when particular selection rules are fulfilled. Such transitions are expected to be enhanced by more than ten orders of magnitude in the presence of the resonator, compared to transitions occurring in homogeneous space. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:26:43 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Almpanis",
"Evangelos",
""
]
] |
2506.04667 | Osayamen Aimuyo | Osayamen Jonathan Aimuyo and Byungsoo Oh and Rachee Singh | FlashDMoE: Fast Distributed MoE in a Single Kernel | In submission. See code at https://github.com/osayamenja/Aristos | null | null | null | cs.DC cs.AR cs.LG | http://creativecommons.org/licenses/by/4.0/ | The computational sparsity of Mixture-of-Experts (MoE) models enables sub-linear growth in compute cost as model size increases, offering a scalable path to training massive neural networks. However, existing implementations suffer from \emph{low GPU utilization}, \emph{significant latency overhead}, and a fundamental \emph{inability to leverage task locality}, primarily due to CPU-managed scheduling, host-initiated communication, and frequent kernel launches. To overcome these limitations, we develop FlashDMoE, a fully GPU-resident MoE operator that fuses expert computation and inter-GPU communication into a \emph{single persistent GPU kernel}. FlashDMoE enables fine-grained pipelining of dispatch, compute, and combine phases, eliminating launch overheads and reducing idle gaps. Its device-initiated communication protocol introduces \emph{payload-efficient} data transfers, significantly shrinking buffer sizes in sparsely activated MoE layers. When evaluated on a single 8-H100 GPU node with MoE models having up to 128 experts and 16K token sequences, FlashDMoE achieves up to \textbf{6}x lower latency, \textbf{5,7}x higher throughput, \textbf{4}x better weak scaling efficiency, and \textbf{9}x higher GPU utilization compared to state-of-the-art baselines, despite using FP32 while baselines use FP16. FlashDMoE demonstrates that principled GPU kernel-hardware co-design is key to unlocking the performance ceiling of large-scale distributed ML workloads. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:29:14 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Aimuyo",
"Osayamen Jonathan",
""
],
[
"Oh",
"Byungsoo",
""
],
[
"Singh",
"Rachee",
""
]
] |
2506.04668 | Takayuki Komatsu | Takayuki Komatsu, Yoshiyuki Ohmura, Kayato Nishitsunoi, and Yasuo Kuniyoshi | Feature-Based Lie Group Transformer for Real-World Applications | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The main goal of representation learning is to acquire meaningful representations from real-world sensory inputs without supervision. Representation learning explains some aspects of human development. Various neural network (NN) models have been proposed that acquire empirically good representations. However, the formulation of a good representation has not been established. We recently proposed a method for categorizing changes between a pair of sensory inputs. A unique feature of this approach is that transformations between two sensory inputs are learned to satisfy algebraic structural constraints. Conventional representation learning often assumes that disentangled independent feature axes is a good representation; however, we found that such a representation cannot account for conditional independence. To overcome this problem, we proposed a new method using group decomposition in Galois algebra theory. Although this method is promising for defining a more general representation, it assumes pixel-to-pixel translation without feature extraction, and can only process low-resolution images with no background, which prevents real-world application. In this study, we provide a simple method to apply our group decomposition theory to a more realistic scenario by combining feature extraction and object segmentation. We replace pixel translation with feature translation and formulate object segmentation as grouping features under the same transformation. We validated the proposed method on a practical dataset containing both real-world object and background. We believe that our model will lead to a better understanding of human development of object recognition in the real world. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:30:11 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Komatsu",
"Takayuki",
""
],
[
"Ohmura",
"Yoshiyuki",
""
],
[
"Nishitsunoi",
"Kayato",
""
],
[
"Kuniyoshi",
"Yasuo",
""
]
] |
2506.04669 | Hanlin Pan | Wanfu Gao, Hanlin Pan, Qingqi Han and Kunpeng Liu | Noise-Resistant Label Reconstruction Feature Selection for Partial Multi-Label Learning | accept in ijcai25 | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The "Curse of dimensionality" is prevalent across various data patterns, which increases the risk of model overfitting and leads to a decline in model classification performance. However, few studies have focused on this issue in Partial Multi-label Learning (PML), where each sample is associated with a set of candidate labels, at least one of which is correct. Existing PML methods addressing this problem are mainly based on the low-rank assumption. However, low-rank assumption is difficult to be satisfied in practical situations and may lead to loss of high-dimensional information. Furthermore, we find that existing methods have poor ability to identify positive labels, which is important in real-world scenarios. In this paper, a PML feature selection method is proposed considering two important characteristics of dataset: label relationship's noise-resistance and label connectivity. Our proposed method utilizes label relationship's noise-resistance to disambiguate labels. Then the learning process is designed through the reformed low-rank assumption. Finally, representative labels are found through label connectivity, and the weight matrix is reconstructed to select features with strong identification ability to these labels. The experimental results on benchmark datasets demonstrate the superiority of the proposed method. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:31:04 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Gao",
"Wanfu",
""
],
[
"Pan",
"Hanlin",
""
],
[
"Han",
"Qingqi",
""
],
[
"Liu",
"Kunpeng",
""
]
] |
2506.04670 | JunJie Huang | Jun-Jie Huang | Geodesic transitive graphs of small valency | null | null | null | null | math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | For a graph $\Gamma$, the {\em distance} $d_\Gamma(u,v)$ between two distinct vertices $u$ and $v$ in $\Gamma$ is defined as the length of the shortest path from $u$ to $v$, and the {\em diameter} $\mathrm{diam}(\Gamma)$ of $\Gamma$ is the maximum distance between $u$ and $v$ for all vertices $u$ and $v$ in the vertex set of $\Gamma$. For a positive integer $s$, a path $(u_0,u_1,\ldots,u_{s})$ is called an {\em $s$-geodesic} if the distance of $u_0$ and $u_s$ is $s$. The graph $\Gamma$ is said to be {\em distance transitive} if for any vertices $u,v,x,y$ of $\Ga$ such that $d_\Ga(u,v)=d_\Ga(x,y)$, there exists an automorphism of $\Gamma$ that maps the pair $(u,v)$ to the pair $(x,y)$. Moreover, $\Gamma$ is said to be {\em geodesic transitive} if for each $i\leq \mathrm{diam}(\Ga)$, the full automorphism group acts transitively on the set of all $i$-geodesics. In the monograph [Distance-Regular Graphs, Section 7.5], the authors listed all distance transitive graphs of valency at most $13$. By using this classification, in this paper, we provide a complete classification of geodesic transitive graphs with valency at most $13$. As a result, there are exactly seven graphs of valency at most $13$ that are distance transitive but not geodesic transitive. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:31:48 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Huang",
"Jun-Jie",
""
]
] |
2506.04671 | Shuren Zhou | Shuren Zhou, Zhao Chen, Yu Yu | CSST Cosmological Emulator III: Hybrid Lagrangian Bias Expansion Emulation of Galaxy Clustering | 29 pages, 12 figures, code at https://github.com/ShurenZhou1999/csstlab | null | null | null | astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | Galaxy clustering is an important probe in the upcoming China Space Station Telescope (CSST) survey to understand the structure growth and reveal the nature of the dark sector. However, it is a long-term challenge to model this biased tracer and connect the observable to the underlying physics. In this work, we present a hybrid Lagrangian bias expansion emulator, combining the Lagrangian bias expansion and the accurate dynamical evolution from $N$-body simulation, to predict the power spectrum of the biased tracer in real space. We employ the Kun simulation suite to construct the emulator, emulating across the space of 8 cosmological parameters including dynamic dark energy $w_0$, $w_a$, and total neutrino mass $\sum m_{\nu}$. The sample variance due to the finite simulation box is further reduced using the Zel'dovich variance control, and it enables the precise measurement of the Lagrangian basis spectra up to quadratic order. The emulation of basis spectra realizes 1% level accuracy, covering wavelength $ k \leq 1 \,{\rm Mpc}^{-1}h$ and redshift $0\leq z\leq 3$ up to quadratic order field. To validate the emulator, we perform the joint fitting of the halo auto power spectrum and the halo-matter cross power spectrum from 46 independent simulations. Depending on the choice of counterpart, the joint fitting is unbiased up to $k_{\rm max}\simeq 0.7\,{\rm Mpc}^{-1}h$ with $1\sim 2$ percent accuracy, for all the redshift and halo mass samples. As one of the CSST cosmological emulator series, our emulator is expected to provide accurate theoretical predictions of the galaxy power spectrum for the upcoming CSST survey. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:36:17 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zhou",
"Shuren",
""
],
[
"Chen",
"Zhao",
""
],
[
"Yu",
"Yu",
""
]
] |
2506.04672 | Shengkun Zhu | Shengkun Zhu, Feiteng Nie, Jinshan Zeng, Sheng Wang, Yuan Sun, Yuan Yao, Shangfeng Chen, Quanqing Xu, Chuanhui Yang | FedAPM: Federated Learning via ADMM with Partial Model Personalization | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | In federated learning (FL), the assumption that datasets from different devices are independent and identically distributed (i.i.d.) often does not hold due to user differences, and the presence of various data modalities across clients makes using a single model impractical. Personalizing certain parts of the model can effectively address these issues by allowing those parts to differ across clients, while the remaining parts serve as a shared model. However, we found that partial model personalization may exacerbate client drift (each client's local model diverges from the shared model), thereby reducing the effectiveness and efficiency of FL algorithms. We propose an FL framework based on the alternating direction method of multipliers (ADMM), referred to as FedAPM, to mitigate client drift. We construct the augmented Lagrangian function by incorporating first-order and second-order proximal terms into the objective, with the second-order term providing fixed correction and the first-order term offering compensatory correction between the local and shared models. Our analysis demonstrates that FedAPM, by using explicit estimates of the Lagrange multiplier, is more stable and efficient in terms of convergence compared to other FL frameworks. We establish the global convergence of FedAPM training from arbitrary initial points to a stationary point, achieving three types of rates: constant, linear, and sublinear, under mild assumptions. We conduct experiments using four heterogeneous and multimodal datasets with different metrics to validate the performance of FedAPM. Specifically, FedAPM achieves faster and more accurate convergence, outperforming the SOTA methods with average improvements of 12.3% in test accuracy, 16.4% in F1 score, and 18.0% in AUC while requiring fewer communication rounds. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:38:29 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zhu",
"Shengkun",
""
],
[
"Nie",
"Feiteng",
""
],
[
"Zeng",
"Jinshan",
""
],
[
"Wang",
"Sheng",
""
],
[
"Sun",
"Yuan",
""
],
[
"Yao",
"Yuan",
""
],
[
"Chen",
"Shangfeng",
""
],
[
"Xu",
"Quanqing",
""
],
[
"Yang",
"Chuanhui",
""
]
] |
2506.04673 | Ronhshuai Wei | Zhong Ji, Rongshuai Wei, Jingren Liu, Yanwei Pang, Jungong Han | Interpretable Few-Shot Image Classification via Prototypical Concept-Guided Mixture of LoRA Experts | 13 pages,5 figures | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Self-Explainable Models (SEMs) rely on Prototypical Concept Learning (PCL) to enable their visual recognition processes more interpretable, but they often struggle in data-scarce settings where insufficient training samples lead to suboptimal performance.To address this limitation, we propose a Few-Shot Prototypical Concept Classification (FSPCC) framework that systematically mitigates two key challenges under low-data regimes: parametric imbalance and representation misalignment. Specifically, our approach leverages a Mixture of LoRA Experts (MoLE) for parameter-efficient adaptation, ensuring a balanced allocation of trainable parameters between the backbone and the PCL module.Meanwhile, cross-module concept guidance enforces tight alignment between the backbone's feature representations and the prototypical concept activation patterns.In addition, we incorporate a multi-level feature preservation strategy that fuses spatial and semantic cues across various layers, thereby enriching the learned representations and mitigating the challenges posed by limited data availability.Finally, to enhance interpretability and minimize concept overlap, we introduce a geometry-aware concept discrimination loss that enforces orthogonality among concepts, encouraging more disentangled and transparent decision boundaries.Experimental results on six popular benchmarks (CUB-200-2011, mini-ImageNet, CIFAR-FS, Stanford Cars, FGVC-Aircraft, and DTD) demonstrate that our approach consistently outperforms existing SEMs by a notable margin, with 4.2%-8.7% relative gains in 5-way 5-shot classification.These findings highlight the efficacy of coupling concept learning with few-shot adaptation to achieve both higher accuracy and clearer model interpretability, paving the way for more transparent visual recognition systems. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:39:43 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Ji",
"Zhong",
""
],
[
"Wei",
"Rongshuai",
""
],
[
"Liu",
"Jingren",
""
],
[
"Pang",
"Yanwei",
""
],
[
"Han",
"Jungong",
""
]
] |
2506.04674 | Jin-Min Liang | Jin-Min Liang, Shao-Ming Fei, Qiongyi He | Variational toolbox-based separability detection of multiqubit states | 9 pages, 4 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Parametrized quantum circuits (PQCs) are crucial in variational quantum algorithms. While it is commonly believed that the optimal PQC is solely used to reproduce the target state, we here reveal that the optimal PQC can also provide valuable insights into the state's properties. We propose variational toolboxes to identify the $k$-separability of pure states, with or without preparation noise, by checking the structure within the optimal PQCs. Additionally, we introduce adaptive optimization strategies to detect the $k$-separability of mixed states. Compared to fixed PQCs, our approach controls fewer parameters for low-rank states. Finally, we validate our methods through numerical demonstrations for various states. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:40:57 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Liang",
"Jin-Min",
""
],
[
"Fei",
"Shao-Ming",
""
],
[
"He",
"Qiongyi",
""
]
] |
2506.04675 | Yui Tomo | Shu Tamano, Yui Tomo | Efficient Gibbs Sampling in Cox Regression Models Using Composite Partial Likelihood and P\'olya-Gamma Augmentation | null | null | null | null | stat.ME | http://creativecommons.org/licenses/by/4.0/ | The Cox regression model and its Bayesian extensions are widely used in survival analysis. However, standard Bayesian approaches require modeling of the baseline hazard, and their full conditional distributions lack closed-form expressions. Therefore, the Metropolis-Hastings sampling algorithm is typically employed, whose efficiency is highly sensitive to the choice of proposal distribution. To address these issues, we propose the GS4Cox, an efficient Gibbs sampling algorithm for the Cox regression model based on four key components: (i) general Bayesian framework, (ii) composite partial likelihood, (iii) P\'olya-Gamma augmentation scheme, and (iv) finite corrections. Our experiments on both synthetic and actual datasets demonstrate that the GS4Cox algorithm outperforms existing sampling methods in terms of convergence speed and sampling efficiency. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:44:23 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Tamano",
"Shu",
""
],
[
"Tomo",
"Yui",
""
]
] |
2506.04676 | I-Sheng Fang | Jing-En Huang, I-Sheng Fang, Tzuhsuan Huang, Chih-Yu Wang, Jun-Cheng Chen | Gen-n-Val: Agentic Image Data Generation and Validation | null | null | null | null | cs.CV cs.AI cs.LG cs.MA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, Large Language Models (LLMs) and Vision Large Language Models (VLLMs) have demonstrated impressive performance as agents across various tasks while data scarcity and label noise remain significant challenges in computer vision tasks, such as object detection and instance segmentation. A common solution for resolving these issues is to generate synthetic data. However, current synthetic data generation methods struggle with issues, such as multiple objects per mask, inaccurate segmentation, and incorrect category labels, limiting their effectiveness. To address these issues, we introduce Gen-n-Val, a novel agentic data generation framework that leverages Layer Diffusion (LD), LLMs, and VLLMs to produce high-quality, single-object masks and diverse backgrounds. Gen-n-Val consists of two agents: (1) The LD prompt agent, an LLM, optimizes prompts for LD to generate high-quality foreground instance images and segmentation masks. These optimized prompts ensure the generation of single-object synthetic data with precise instance masks and clean backgrounds. (2) The data validation agent, a VLLM, which filters out low-quality synthetic instance images. The system prompts for both agents are refined through TextGrad. Additionally, we use image harmonization to combine multiple instances within scenes. Compared to state-of-the-art synthetic data approaches like MosaicFusion, our approach reduces invalid synthetic data from 50% to 7% and improves performance by 1% mAP on rare classes in COCO instance segmentation with YOLOv9c and YOLO11m. Furthermore, Gen-n-Val shows significant improvements (7. 1% mAP) over YOLO-Worldv2-M in open-vocabulary object detection benchmarks with YOLO11m. Moreover, Gen-n-Val improves the performance of YOLOv9 and YOLO11 families in instance segmentation and object detection. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:52:26 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Huang",
"Jing-En",
""
],
[
"Fang",
"I-Sheng",
""
],
[
"Huang",
"Tzuhsuan",
""
],
[
"Wang",
"Chih-Yu",
""
],
[
"Chen",
"Jun-Cheng",
""
]
] |
2506.04677 | Marco Zanotti | Marco Zanotti | The cost of ensembling: is it always worth combining? | null | null | null | null | cs.LG stat.AP stat.OT | http://creativecommons.org/licenses/by/4.0/ | Given the continuous increase in dataset sizes and the complexity of forecasting models, the trade-off between forecast accuracy and computational cost is emerging as an extremely relevant topic, especially in the context of ensemble learning for time series forecasting. To asses it, we evaluated ten base models and eight ensemble configurations across two large-scale retail datasets (M5 and VN1), considering both point and probabilistic accuracy under varying retraining frequencies. We showed that ensembles consistently improve forecasting performance, particularly in probabilistic settings. However, these gains come at a substantial computational cost, especially for larger, accuracy-driven ensembles. We found that reducing retraining frequency significantly lowers costs, with minimal impact on accuracy, particularly for point forecasts. Moreover, efficiency-driven ensembles offer a strong balance, achieving competitive accuracy with considerably lower costs compared to accuracy-optimized combinations. Most importantly, small ensembles of two or three models are often sufficient to achieve near-optimal results. These findings provide practical guidelines for deploying scalable and cost-efficient forecasting systems, supporting the broader goals of sustainable AI in forecasting. Overall, this work shows that careful ensemble design and retraining strategy selection can yield accurate, robust, and cost-effective forecasts suitable for real-world applications. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:54:19 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zanotti",
"Marco",
""
]
] |
2506.04678 | Wendi Cheng | Ming Li, Wendi Cheng, Jiahe Wei, Xueqiang Shan, Liu Weikai, Xiaonan Zhao, Xiao Zhang | BVLSM: Write-Efficient LSM-Tree Storage via WAL-Time Key-Value Separation | null | null | null | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Modern data-intensive applications increasingly store and process big-value items, such as multimedia objects and machine learning embeddings, which exacerbate storage inefficiencies in Log-Structured Merge-Tree (LSM)-based key-value stores. This paper presents BVLSM, a Write-Ahead Log (WAL)-time key-value separation mechanism designed to address three key challenges in LSM-Tree storage systems: write amplification, poor memory utilization, and I/O jitter under big-value workloads. Unlike state-of-the-art approaches that delay key-value separation until the flush stage, leading to redundant data in MemTables and repeated writes. BVLSM proactively decouples keys and values during the WAL phase. The MemTable stores only lightweight metadata, allowing multi-queue parallel store for big value. The benchmark results show that BVLSM significantly outperforms both RocksDB and BlobDB under 64KB random write workloads. In asynchronous WAL mode, it achieves throughput improvements of 7.6x over RocksDB and 1.9x over BlobDB. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:54:25 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Li",
"Ming",
""
],
[
"Cheng",
"Wendi",
""
],
[
"Wei",
"Jiahe",
""
],
[
"Shan",
"Xueqiang",
""
],
[
"Weikai",
"Liu",
""
],
[
"Zhao",
"Xiaonan",
""
],
[
"Zhang",
"Xiao",
""
]
] |
2506.04679 | Rapha\"el Milli\`ere | Rapha\"el Milli\`ere | Normative Conflicts and Shallow AI Alignment | Published in Philosophical Studies | Milliere, R. (2025). Normative conflicts and shallow AI alignment. Philosophical Studies, 1-44 | 10.1007/s11098-025-02347-3 | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | The progress of AI systems such as large language models (LLMs) raises increasingly pressing concerns about their safe deployment. This paper examines the value alignment problem for LLMs, arguing that current alignment strategies are fundamentally inadequate to prevent misuse. Despite ongoing efforts to instill norms such as helpfulness, honesty, and harmlessness in LLMs through fine-tuning based on human preferences, they remain vulnerable to adversarial attacks that exploit conflicts between these norms. I argue that this vulnerability reflects a fundamental limitation of existing alignment methods: they reinforce shallow behavioral dispositions rather than endowing LLMs with a genuine capacity for normative deliberation. Drawing from on research in moral psychology, I show how humans' ability to engage in deliberative reasoning enhances their resilience against similar adversarial tactics. LLMs, by contrast, lack a robust capacity to detect and rationally resolve normative conflicts, leaving them susceptible to manipulation; even recent advances in reasoning-focused LLMs have not addressed this vulnerability. This ``shallow alignment'' problem carries significant implications for AI safety and regulation, suggesting that current approaches are insufficient for mitigating potential harms posed by increasingly capable AI systems. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 06:57:28 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Millière",
"Raphaël",
""
]
] |
2506.04680 | PingKong Hunag | Ping-Kong Huang, Chien-Wu Lan, Chin-Tien Wu | Application of SDRE to Achieve Gait Control in a Bipedal Robot for Knee-Type Exoskeleton Testing | 8 pages, 6 figures. Preliminary version submitted for documentation purposes on arXiv. This version records results presented at a conference and is not peer-reviewed | null | null | null | cs.RO math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Exoskeletons are widely used in rehabilitation and industrial applications to assist human motion. However, direct human testing poses risks due to possible exoskeleton malfunctions and inconsistent movement replication. To provide a safer and more repeatable testing environment, this study employs a bipedal robot platform to reproduce human gait, allowing for controlled exoskeleton evaluations. A control strategy based on the State-Dependent Riccati Equation (SDRE) is formulated to achieve optimal torque control for accurate gait replication. The bipedal robot dynamics are represented using double pendulum model, where SDRE-optimized control inputs minimize deviations from human motion trajectories. To align with motor behavior constraints, a parameterized control method is introduced to simplify the control process while effectively replicating human gait. The proposed approach initially adopts a ramping trapezoidal velocity model, which is then adapted into a piecewise linear velocity-time representation through motor command overwriting. This modification enables finer control over gait phase transitions while ensuring compatibility with motor dynamics. The corresponding cost function optimizes the control parameters to minimize errors in joint angles, velocities, and torques relative to SDRE control result. By structuring velocity transitions in accordance with motor limitations, the method reduce the computational load associated with real-time control. Experimental results verify the feasibility of the proposed parameterized control method in reproducing human gait. The bipedal robot platform provides a reliable and repeatable testing mechanism for knee-type exoskeletons, offering insights into exoskeleton performance under controlled conditions. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:00:10 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Huang",
"Ping-Kong",
""
],
[
"Lan",
"Chien-Wu",
""
],
[
"Wu",
"Chin-Tien",
""
]
] |
2506.04681 | Daogao Liu | Daogao Liu, Edith Cohen, Badih Ghazi, Peter Kairouz, Pritish Kamath, Alexander Knop, Ravi Kumar, Pasin Manurangsi, Adam Sealfon, Da Yu, Chiyuan Zhang | Urania: Differentially Private Insights into AI Use | null | null | null | null | cs.LG cs.AI cs.CL cs.CR cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce $Urania$, a novel framework for generating insights about LLM chatbot interactions with rigorous differential privacy (DP) guarantees. The framework employs a private clustering mechanism and innovative keyword extraction methods, including frequency-based, TF-IDF-based, and LLM-guided approaches. By leveraging DP tools such as clustering, partition selection, and histogram-based summarization, $Urania$ provides end-to-end privacy protection. Our evaluation assesses lexical and semantic content preservation, pair similarity, and LLM-based metrics, benchmarking against a non-private Clio-inspired pipeline (Tamkin et al., 2024). Moreover, we develop a simple empirical privacy evaluation that demonstrates the enhanced robustness of our DP pipeline. The results show the framework's ability to extract meaningful conversational insights while maintaining stringent user privacy, effectively balancing data utility with privacy preservation. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:00:31 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Liu",
"Daogao",
""
],
[
"Cohen",
"Edith",
""
],
[
"Ghazi",
"Badih",
""
],
[
"Kairouz",
"Peter",
""
],
[
"Kamath",
"Pritish",
""
],
[
"Knop",
"Alexander",
""
],
[
"Kumar",
"Ravi",
""
],
[
"Manurangsi",
"Pasin",
""
],
[
"Sealfon",
"Adam",
""
],
[
"Yu",
"Da",
""
],
[
"Zhang",
"Chiyuan",
""
]
] |
2506.04682 | Chuyun Deng | Chuyun Deng, Na Liu, Wei Xie, Lianming Xu and Li Wang | MARS: Radio Map Super-resolution and Reconstruction Method under Sparse Channel Measurements | null | null | null | null | cs.CV eess.SP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Radio maps reflect the spatial distribution of signal strength and are essential for applications like smart cities, IoT, and wireless network planning. However, reconstructing accurate radio maps from sparse measurements remains challenging. Traditional interpolation and inpainting methods lack environmental awareness, while many deep learning approaches depend on detailed scene data, limiting generalization. To address this, we propose MARS, a Multi-scale Aware Radiomap Super-resolution method that combines CNNs and Transformers with multi-scale feature fusion and residual connections. MARS focuses on both global and local feature extraction, enhancing feature representation across different receptive fields and improving reconstruction accuracy. Experiments across different scenes and antenna locations show that MARS outperforms baseline models in both MSE and SSIM, while maintaining low computational cost, demonstrating strong practical potential. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:03:50 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Deng",
"Chuyun",
""
],
[
"Liu",
"Na",
""
],
[
"Xie",
"Wei",
""
],
[
"Xu",
"Lianming",
""
],
[
"Wang",
"Li",
""
]
] |
2506.04683 | Jitendra Singh Mr | Jitendra Singh, Anand Mehrotra, Suraj Srivastava, Aditya K. Jagannatham, and Lajos Hanzo | Spectral Efficiency Maximization for mmWave MIMO-Aided Integrated Sensing and Communication Under Practical Constraints | null | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | A hybrid transmit precoder (TPC) and receive combiner (RC) pair is conceived for millimeter wave (mmWave) multiple input multiple output (MIMO) integrated sensing and communication (ISAC) systems. The proposed design considers a practical mean squared error (MSE) constraint between the desired and the achieved beampatterns constructed for identifying radar targets (RTs). To achieve optimal performance, we formulate an optimization problem relying on sum spectral efficiency (SE) maximization of the communication users (CUs), while satisfying certain radar beampattern similarity (RBPS), total transmit power, and constant modulus constraints, where the latter are attributed to the hybrid mmWave MIMO architecture. Since the aforementioned problem is non-convex and intractable, a sequential approach is proposed wherein the TPCs are designed first, followed by the RCs. To deal with the non-convex MSE and constant modulus constraints in the TPC design problem, we propose a majorization and minimization (MM) based Riemannian conjugate gradient (RCG) method, which restricts the tolerable MSE of the beampattern to within a predefined limit. Moreover, the least squares and the zero-forcing methods are adopted for maximizing the sum-SE and for mitigating the multiuser interference (MUI), respectively. Furthermore, to design the RC at each CU, we propose a linear MM-based blind combiner (LMBC) scheme that does not rely on the knowledge of the TPC at the CUs and has a low complexity. To achieve user fairness, we further extend the proposed sequential approach for maximizing the geometric mean (GM) of the CU's rate. Simulation results are presented, which show the superior performance of the proposed hybrid TPC and RC in comparison to the state-of-the-art designs in the mmWave MIMO ISAC systems under consideration. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:03:52 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Singh",
"Jitendra",
""
],
[
"Mehrotra",
"Anand",
""
],
[
"Srivastava",
"Suraj",
""
],
[
"Jagannatham",
"Aditya K.",
""
],
[
"Hanzo",
"Lajos",
""
]
] |
2506.04684 | Nitish Kumar Mr. | Nitish Kumar and Rajalakshmi Pachamuthu | Real-Time LPV-Based Non-Linear Model Predictive Control for Robust Trajectory Tracking in Autonomous Vehicles | null | null | null | null | cs.RO cs.SY eess.SY | http://creativecommons.org/licenses/by/4.0/ | This paper presents the development and implementation of a Model Predictive Control (MPC) framework for trajectory tracking in autonomous vehicles under diverse driving conditions. The proposed approach incorporates a modular architecture that integrates state estimation, vehicle dynamics modeling, and optimization to ensure real-time performance. The state-space equations are formulated in a Linear Parameter Varying (LPV) form, and a curvature-based tuning method is introduced to optimize weight matrices for varying trajectories. The MPC framework is implemented using the Robot Operating System (ROS) for parallel execution of state estimation and control optimization, ensuring scalability and minimal latency. Extensive simulations and real-time experiments were conducted on multiple predefined trajectories, demonstrating high accuracy with minimal cross-track and orientation errors, even under aggressive maneuvers and high-speed conditions. The results highlight the robustness and adaptability of the proposed system, achieving seamless alignment between simulated and real-world performance. This work lays the foundation for dynamic weight tuning and integration into cooperative autonomous navigation systems, paving the way for enhanced safety and efficiency in autonomous driving applications. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:04:10 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Kumar",
"Nitish",
""
],
[
"Pachamuthu",
"Rajalakshmi",
""
]
] |
2506.04685 | Andreas Hadjigeorgiou | Andreas Hadjigeorgiou and Stelios Timotheou | Energy Consumption Optimization for Autonomous Vehicles via Positive Control Input Minimization | null | null | null | null | math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Autonomous vehicles (AVs) present a unique opportunity to improve the sustainability of transportation systems by adopting eco-driving strategies that reduce energy consumption and emissions. This paper introduces a novel surrogate model for energy and fuel consumption that minimizes Positive Control Input (PCI). Unlike conventional objectives such as squared acceleration, which often misrepresent actual energy usage, PCI provides a more accurate and optimization-friendly alternative. Building on PCI, we propose ECO+, a convex, time-based trajectory optimization framework that ensures safety and passenger comfort while optimizing energy use for AVs approaching an intersection. To improve computational efficiency, quadratic resistive forces are approximated using piecewise affine segments, resulting in a linear programming formulation. ECO+ is validated using empirical fuel and electric energy models and benchmarked against established optimization strategies, including a state-of-the-art nonlinear solver. Simulation results demonstrate that ECO+ consistently outperforms baseline methods in reducing energy consumption, even under strict comfort constraints and in scenarios involving a leading vehicle. Moreover, initializing a nonlinear solver with ECO+ yields only marginal gains, indicating that ECO+ is effective as a standalone eco-driving strategy. These findings highlight ECO+ as a practical, scalable, and computationally efficient solution for enhancing the sustainability of autonomous urban mobility systems. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:06:16 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Hadjigeorgiou",
"Andreas",
""
],
[
"Timotheou",
"Stelios",
""
]
] |
2506.04686 | Nicolas Borchard | Nicolas Borchard and Gerd Wachsmuth | Characterization of Hilbertizable spaces via convex functions | 8 pages | null | null | null | math.FA | http://creativecommons.org/licenses/by-sa/4.0/ | We show that the existence of a strongly convex function with a Lipschitz derivative on a Banach space already implies that the space is isomorphic to a Hilbert space. Similarly, if both a function and its convex conjugate are $C^2$ then the underlying space is also isomorphic to a Hilbert space. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:06:28 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Borchard",
"Nicolas",
""
],
[
"Wachsmuth",
"Gerd",
""
]
] |
2506.04687 | Akihisa Okada | Akihisa Okada, Keisuke Otaki, and Hiroaki Yoshida | Joint Optimization of Electric Vehicle Routes and Charging Locations Learning Charge Constraints Using QUBO Solver | 12 pages, 10 figures | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Optimal routing problems of electric vehicles (EVs) have attracted much attention in recent years, and installation of charging stations is an important issue for EVs. Hence, we focus on the joint optimization of the location of charging stations and the routing of EVs. When routing problems are formulated in the form of quadratic unconstrained binary optimization (QUBO), specialized solvers such as quantum annealer are expected to provide optimal solutions with high speed and accuracy. However, battery capacity constraints make it hard to formulate into QUBO form without a large number of auxiliary qubits. Then, we propose a sequential optimization method utilizing the Bayesian inference and QUBO solvers, in which method the battery capacity constraints are automatically learned. This method enables us to optimize the number and location of charging stations and the routing of EVs with a small number of searches. Applying this method to a routing problem of 20 locations, we confirmed that the learning process works well and efficient searches find good solutions. This result enhances the possibility that the QUBO solver could be applied to the constraints contained problems which is difficult to formulate into QUBO form without a large number of ancilla qubits. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:08:19 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Okada",
"Akihisa",
""
],
[
"Otaki",
"Keisuke",
""
],
[
"Yoshida",
"Hiroaki",
""
]
] |
2506.04688 | Gio Paik | Gio Paik, Geewook Kim, Jinbae Im | MMRefine: Unveiling the Obstacles to Robust Refinement in Multimodal Large Language Models | ACL Findings 2025 | null | null | null | cs.CL cs.AI cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper introduces MMRefine, a MultiModal Refinement benchmark designed to evaluate the error refinement capabilities of Multimodal Large Language Models (MLLMs). As the emphasis shifts toward enhancing reasoning during inference, MMRefine provides a framework that evaluates MLLMs' abilities to detect and correct errors across six distinct scenarios beyond just comparing final accuracy before and after refinement. Furthermore, the benchmark analyzes the refinement performance by categorizing errors into six error types. Experiments with various open and closed MLLMs reveal bottlenecks and factors impeding refinement performance, highlighting areas for improvement in effective reasoning enhancement. Our code and dataset are publicly available at https://github.com/naver-ai/MMRefine. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:11:36 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Paik",
"Gio",
""
],
[
"Kim",
"Geewook",
""
],
[
"Im",
"Jinbae",
""
]
] |
2506.04689 | Thao Nguyen | Thao Nguyen, Yang Li, Olga Golovneva, Luke Zettlemoyer, Sewoong Oh, Ludwig Schmidt, Xian Li | Recycling the Web: A Method to Enhance Pre-training Data Quality and Quantity for Language Models | null | null | null | null | cs.CL cs.LG | http://creativecommons.org/licenses/by/4.0/ | Scaling laws predict that the performance of large language models improves with increasing model size and data size. In practice, pre-training has been relying on massive web crawls, using almost all data sources publicly available on the internet so far. However, this pool of natural data does not grow at the same rate as the compute supply. Furthermore, the availability of high-quality texts is even more limited: data filtering pipelines often remove up to 99% of the initial web scrapes to achieve state-of-the-art. To address the "data wall" of pre-training scaling, our work explores ways to transform and recycle data discarded in existing filtering processes. We propose REWIRE, REcycling the Web with guIded REwrite, a method to enrich low-quality documents so that they could become useful for training. This in turn allows us to increase the representation of synthetic data in the final pre-training set. Experiments at 1B, 3B and 7B scales of the DCLM benchmark show that mixing high-quality raw texts and our rewritten texts lead to 1.0, 1.3 and 2.5 percentage points improvement respectively across 22 diverse tasks, compared to training on only filtered web data. Training on the raw-synthetic data mix is also more effective than having access to 2x web data. Through further analysis, we demonstrate that about 82% of the mixed in texts come from transforming lower-quality documents that would otherwise be discarded. REWIRE also outperforms related approaches of generating synthetic data, including Wikipedia-style paraphrasing, question-answer synthesizing and knowledge extraction. These results suggest that recycling web texts holds the potential for being a simple and effective approach for scaling pre-training data. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:12:12 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Nguyen",
"Thao",
""
],
[
"Li",
"Yang",
""
],
[
"Golovneva",
"Olga",
""
],
[
"Zettlemoyer",
"Luke",
""
],
[
"Oh",
"Sewoong",
""
],
[
"Schmidt",
"Ludwig",
""
],
[
"Li",
"Xian",
""
]
] |
2506.04690 | Yifan Hao | Yifan Hao, Yanxin Lu, Xinwei Shen, Tong Zhang | Towards Better Generalization via Distributional Input Projection Network | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As overparameterized models become increasingly prevalent, training loss alone offers limited insight into generalization performance. While smoothness has been linked to improved generalization across various settings, directly enforcing smoothness in neural networks remains challenging. To address this, we introduce Distributional Input Projection Networks (DIPNet), a novel framework that projects inputs into learnable distributions at each layer. This distributional representation induces a smoother loss landscape with respect to the input, promoting better generalization. We provide theoretical analysis showing that DIPNet reduces both local smoothness measures and the Lipschitz constant of the network, contributing to improved generalization performance. Empirically, we validate DIPNet across a wide range of architectures and tasks, including Vision Transformers (ViTs), Large Language Models (LLMs), ResNet and MLPs. Our method consistently enhances test performance under standard settings, adversarial attacks, out-of-distribution inputs, and reasoning benchmarks. We demonstrate that the proposed input projection strategy can be seamlessly integrated into existing models, providing a general and effective approach for boosting generalization performance in modern deep learning. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:13:59 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Hao",
"Yifan",
""
],
[
"Lu",
"Yanxin",
""
],
[
"Shen",
"Xinwei",
""
],
[
"Zhang",
"Tong",
""
]
] |
2506.04691 | Pascal Begout | Pascal B\'egout (IMT), Jesus Ildefonso Diaz | Solutions with expanding compact support of saturated Schr{\"o}dinger equations: self-similar solutions | null | Electronic Journal of Differential Equations, 2025, 2025 (53), pp.01-15 | null | null | math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We prove the existence of solutions \(u(t,x)\) of the Schr{\"o}dinger equation with a saturation nonlinear term \((u/|u|)\) having compact support, for each \(t>0,\) that expands with a growth law of the type \(C\sqrt{t}\). The primary tool is considering the self-similar solution of the associated equation. For more information see https://ejde.math.txstate.edu/Volumes/2025/53/abstr.html | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:14:50 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Bégout",
"Pascal",
"",
"IMT"
],
[
"Diaz",
"Jesus Ildefonso",
""
]
] |
2506.04692 | Lorenzo Luperi Baglini | Leonardo Raffaello Maximilian Gasparro, Lorenzo Luperi Baglini | Nonstandard methods for ultrafilter relation extensions | null | null | null | null | math.LO | http://creativecommons.org/licenses/by/4.0/ | We study two types of ultrafilter relation extensions with nonstandard methods and provide characterisations of these extensions in terms of generators of ultrafilters. These characterisations are then used to determine whether various relation properties are preserved by these extensions. Lastly, we study the notion of self-divisible ultrafilter and, with the mentioned characterizations, generalize this notion for a class of relations. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:15:10 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Gasparro",
"Leonardo Raffaello Maximilian",
""
],
[
"Baglini",
"Lorenzo Luperi",
""
]
] |
2506.04693 | Lu Wei | Lu Wei, Liangzhi Li, Tong Xiang, Xiao Liu, Noa Garcia | Cracking the Code: Enhancing Implicit Hate Speech Detection through Coding Classification | Proceedings of the 5th Workshop on Trustworthy NLP (TrustNLP 2025), 112-126 | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | The internet has become a hotspot for hate speech (HS), threatening societal harmony and individual well-being. While automatic detection methods perform well in identifying explicit hate speech (ex-HS), they struggle with more subtle forms, such as implicit hate speech (im-HS). We tackle this problem by introducing a new taxonomy for im-HS detection, defining six encoding strategies named codetypes. We present two methods for integrating codetypes into im-HS detection: 1) prompting large language models (LLMs) directly to classify sentences based on generated responses, and 2) using LLMs as encoders with codetypes embedded during the encoding process. Experiments show that the use of codetypes improves im-HS detection in both Chinese and English datasets, validating the effectiveness of our approach across different languages. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:15:21 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Wei",
"Lu",
""
],
[
"Li",
"Liangzhi",
""
],
[
"Xiang",
"Tong",
""
],
[
"Liu",
"Xiao",
""
],
[
"Garcia",
"Noa",
""
]
] |
2506.04694 | Jaeseung Heo | Jaeseung Heo, Kyeongheung Yun, Seokwon Yoon, MoonJeong Park, Jungseul Ok, Dongwoo Kim | Influence Functions for Edge Edits in Non-Convex Graph Neural Networks | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Understanding how individual edges influence the behavior of graph neural networks (GNNs) is essential for improving their interpretability and robustness. Graph influence functions have emerged as promising tools to efficiently estimate the effects of edge deletions without retraining. However, existing influence prediction methods rely on strict convexity assumptions, exclusively consider the influence of edge deletions while disregarding edge insertions, and fail to capture changes in message propagation caused by these modifications. In this work, we propose a proximal Bregman response function specifically tailored for GNNs, relaxing the convexity requirement and enabling accurate influence prediction for standard neural network architectures. Furthermore, our method explicitly accounts for message propagation effects and extends influence prediction to both edge deletions and insertions in a principled way. Experiments with real-world datasets demonstrate accurate influence predictions for different characteristics of GNNs. We further demonstrate that the influence function is versatile in applications such as graph rewiring and adversarial attacks. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:15:46 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Heo",
"Jaeseung",
""
],
[
"Yun",
"Kyeongheung",
""
],
[
"Yoon",
"Seokwon",
""
],
[
"Park",
"MoonJeong",
""
],
[
"Ok",
"Jungseul",
""
],
[
"Kim",
"Dongwoo",
""
]
] |
2506.04695 | Difan Zou | Xingwu Chen, Tianle Li, Difan Zou | On the Mechanism of Reasoning Pattern Selection in Reinforcement Learning for Language Models | 30 pages, 6 figures, 1 table | null | null | null | cs.LG cs.AI stat.ML | http://creativecommons.org/licenses/by/4.0/ | Reinforcement learning (RL) has demonstrated remarkable success in enhancing model capabilities, including instruction-following, preference learning, and reasoning. Yet despite its empirical successes, the mechanisms by which RL improves reasoning abilities remain poorly understood. We present a systematic study of Reinforcement Learning with Verifiable Rewards (RLVR), showing that its primary benefit comes from optimizing the selection of existing reasoning patterns. Through extensive experiments, we demonstrate that RLVR-trained models preferentially adopt high-success-rate reasoning patterns while mostly maintaining stable performance on individual patterns. We further develop theoretical analyses on the convergence and training dynamics of RLVR based on a simplified question-reason-answer model. We study the gradient flow and show that RLVR can indeed find the solution that selects the reason pattern with the highest success rate. Besides, our theoretical results
reveal two distinct regimes regarding the convergence of RLVR training: (1) rapid convergence for models with relatively strong initial reasoning capabilities versus (2) slower optimization dynamics for weaker models. Furthermore, we show that the slower optimization for weaker models can be mitigated by applying the supervised fine-tuning (SFT) before RLVR, when using a feasibly high-quality SFT dataset. We validate the theoretical findings through extensive experiments. This work advances our theoretical understanding of RL's role in LLM fine-tuning and offers insights for further enhancing reasoning capabilities. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:17:04 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Chen",
"Xingwu",
""
],
[
"Li",
"Tianle",
""
],
[
"Zou",
"Difan",
""
]
] |
2506.04696 | Tonmoy Paul | Tonmoy Paul, Mrittika Devi Mati and Md. Mahmudul Islam | Enhanced Drought Analysis in Bangladesh: A Machine Learning Approach for Severity Classification Using Satellite Data | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Drought poses a pervasive environmental challenge in Bangladesh, impacting agriculture, socio-economic stability, and food security due to its unique geographic and anthropogenic vulnerabilities. Traditional drought indices, such as the Standardized Precipitation Index (SPI) and Palmer Drought Severity Index (PDSI), often overlook crucial factors like soil moisture and temperature, limiting their resolution. Moreover, current machine learning models applied to drought prediction have been underexplored in the context of Bangladesh, lacking a comprehensive integration of satellite data across multiple districts. To address these gaps, we propose a satellite data-driven machine learning framework to classify drought across 38 districts of Bangladesh. Using unsupervised algorithms like K-means and Bayesian Gaussian Mixture for clustering, followed by classification models such as KNN, Random Forest, Decision Tree, and Naive Bayes, the framework integrates weather data (humidity, soil moisture, temperature) from 2012-2024. This approach successfully classifies drought severity into different levels. However, it shows significant variabilities in drought vulnerabilities across regions which highlights the aptitude of machine learning models in terms of identifying and predicting drought conditions. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:17:43 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Paul",
"Tonmoy",
""
],
[
"Mati",
"Mrittika Devi",
""
],
[
"Islam",
"Md. Mahmudul",
""
]
] |
2506.04697 | Yingying Peng | Xinyi Jiang, Qizhi Li, Qingzheng Qiu, Li Yue, Junhan Huang, Yiwen Chen, Byungjune Lee, Hyeongi Choi, Xingjiang Zhou, Tao Dong, Nanlin Wang, Hoyoung Jang, and Yingying Peng | Ultrafast Orbital-Selective Photodoping Melts Charge Order in Overdoped Bi-based Cuprates | 13 pages, 11 figures, comments are welcome | null | null | null | cond-mat.supr-con cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | High-temperature superconductivity in cuprates remains one of the enduring puzzles of condensed matter physics, with charge order (CO) playing a central yet elusive role, particularly in the overdoped regime. Here, we employ time-resolved X-ray absorption spectroscopy and resonant X-ray scattering at a free-electron laser to probe the transient electronic density of states and ultrafast CO dynamics in overdoped (Bi,Pb)$_{2.12}$Sr$_{1.88}$CuO$_{6+\delta}$. We reveal a striking pump laser wavelength dependence - the 800 nm light fails to suppress CO, whereas the 400 nm light effectively melts it. This behavior originates from the fact that 400 nm photons can promote electrons from the Zhang-Rice singlet band to the upper Hubbard band or apical oxygen states, while 800 nm photons lack the energy to excite electrons across the charge-transfer gap. The CO recovery time ($\sim$3 ps) matches that of the underdoped cuprates, indicating universal electronic instability in the phase diagram. Additionally, melting overdoped CO requires an order-of-magnitude higher fluence highlighting the role of lattice interactions. Our findings demonstrate orbital-selective photodoping and provide a route to ultrafast control of emergent quantum phases in correlated materials. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:18:31 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Jiang",
"Xinyi",
""
],
[
"Li",
"Qizhi",
""
],
[
"Qiu",
"Qingzheng",
""
],
[
"Yue",
"Li",
""
],
[
"Huang",
"Junhan",
""
],
[
"Chen",
"Yiwen",
""
],
[
"Lee",
"Byungjune",
""
],
[
"Choi",
"Hyeongi",
""
],
[
"Zhou",
"Xingjiang",
""
],
[
"Dong",
"Tao",
""
],
[
"Wang",
"Nanlin",
""
],
[
"Jang",
"Hoyoung",
""
],
[
"Peng",
"Yingying",
""
]
] |
2506.04698 | Michail-Antisthenis Tsompanas | Hugo Alcaraz-Herrera, Michail-Antisthenis Tsompanas, Igor Balaz, Andrew Adamatzky | NEAT and HyperNEAT based Design for Soft Actuator Controllers | null | null | null | null | cs.NE | http://creativecommons.org/licenses/by-sa/4.0/ | Since soft robotics are composed of compliant materials, they perform better than conventional rigid robotics in specific fields, such as medical applications. However, the field of soft robotics is fairly new, and the design process of their morphology and their controller strategies has not yet been thoroughly studied. Consequently, here, an automated design method for the controller of soft actuators based on Neuroevolution is proposed. Specifically, the suggested techniques employ Neuroevolution of Augmenting Topologies (NEAT) and Hypercube-based NEAT (HyperNEAT) to generate the synchronization profile of the components of a simulated soft actuator by employing Compositional Pattern Producing Networks (CPPNs). As a baseline methodology, a Standard Genetic Algorithm (SGA) was used. Moreover, to test the robustness of the proposed methodologies, both high- and low-performing morphologies of soft actuators were utilized as testbeds. Moreover, the use of an affluent and a more limited set of activation functions for the Neuroevolution targets was tested throughout the experiments. The results support the hypothesis that Neuroevolution based methodologies are more appropriate for designing controllers that align with both types of morphologies. In specific, NEAT performed better for all different scenarios tested and produced more simplistic networks that are easier to implement in real life applications. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:20:16 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Alcaraz-Herrera",
"Hugo",
""
],
[
"Tsompanas",
"Michail-Antisthenis",
""
],
[
"Balaz",
"Igor",
""
],
[
"Adamatzky",
"Andrew",
""
]
] |
2506.04699 | Bihan Xu | Bihan Xu, Shiwei Zhao, Runze Wu, Zhenya Huang, Jiawei Wang, Zhipeng Hu, Kai Wang, Haoyu Liu, Tangjie Lv, Le Li, Changjie Fan, Xin Tong, Jiangze Han | Empowering Economic Simulation for Massively Multiplayer Online Games through Generative Agent-Based Modeling | KDD2025 Accepted | null | null | null | cs.AI | http://creativecommons.org/licenses/by-sa/4.0/ | Within the domain of Massively Multiplayer Online (MMO) economy research, Agent-Based Modeling (ABM) has emerged as a robust tool for analyzing game economics, evolving from rule-based agents to decision-making agents enhanced by reinforcement learning. Nevertheless, existing works encounter significant challenges when attempting to emulate human-like economic activities among agents, particularly regarding agent reliability, sociability, and interpretability. In this study, we take a preliminary step in introducing a novel approach using Large Language Models (LLMs) in MMO economy simulation. Leveraging LLMs' role-playing proficiency, generative capacity, and reasoning aptitude, we design LLM-driven agents with human-like decision-making and adaptability. These agents are equipped with the abilities of role-playing, perception, memory, and reasoning, addressing the aforementioned challenges effectively. Simulation experiments focusing on in-game economic activities demonstrate that LLM-empowered agents can promote emergent phenomena like role specialization and price fluctuations in line with market rules. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:21:13 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Xu",
"Bihan",
""
],
[
"Zhao",
"Shiwei",
""
],
[
"Wu",
"Runze",
""
],
[
"Huang",
"Zhenya",
""
],
[
"Wang",
"Jiawei",
""
],
[
"Hu",
"Zhipeng",
""
],
[
"Wang",
"Kai",
""
],
[
"Liu",
"Haoyu",
""
],
[
"Lv",
"Tangjie",
""
],
[
"Li",
"Le",
""
],
[
"Fan",
"Changjie",
""
],
[
"Tong",
"Xin",
""
],
[
"Han",
"Jiangze",
""
]
] |
2506.04700 | Jos\'e Manuel De Frutos | Jos\'e Manuel de Frutos, Manuel A. V\'azquez, Pablo M. Olmos, Joaqu\'in M\'iguez | Explicit Density Approximation for Neural Implicit Samplers Using a Bernstein-Based Convex Divergence | null | null | null | null | cs.LG cs.AI math.PR stat.ML | http://creativecommons.org/licenses/by/4.0/ | Rank-based statistical metrics, such as the invariant statistical loss (ISL), have recently emerged as robust and practically effective tools for training implicit generative models. In this work, we introduce dual-ISL, a novel likelihood-free objective for training implicit generative models that interchanges the roles of the target and model distributions in the ISL framework, yielding a convex optimization problem in the space of model densities. We prove that the resulting rank-based discrepancy $d_K$ is i) continuous under weak convergence and with respect to the $L^1$ norm, and ii) convex in its first argument-properties not shared by classical divergences such as KL or Wasserstein distances. Building on this, we develop a theoretical framework that interprets $d_K$ as an $L^2$-projection of the density ratio $q = p/\tilde p$ onto a Bernstein polynomial basis, from which we derive exact bounds on the truncation error, precise convergence rates, and a closed-form expression for the truncated density approximation. We further extend our analysis to the multivariate setting via random one-dimensional projections, defining a sliced dual-ISL divergence that retains both convexity and continuity. We empirically show that these theoretical advantages translate into practical ones. Specifically, across several benchmarks dual-ISL converges more rapidly, delivers markedly smoother and more stable training, and more effectively prevents mode collapse than classical ISL and other leading implicit generative methods-while also providing an explicit density approximation. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:21:54 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"de Frutos",
"José Manuel",
""
],
[
"Vázquez",
"Manuel A.",
""
],
[
"Olmos",
"Pablo M.",
""
],
[
"Míguez",
"Joaquín",
""
]
] |
2506.04701 | Wei Su | Meiru Jiang, Wei Su, Guojian Ren, Yongguang Yu | Memory-Driven Bounded Confidence Opinion Dynamics: A Hegselmann-Krause Model Based on Fractional-Order Methods | null | null | null | null | physics.soc-ph cs.MA cs.SI nlin.AO | http://creativecommons.org/licenses/by/4.0/ | Memory effects play a crucial role in social interactions and decision-making processes. This paper proposes a novel fractional-order bounded confidence opinion dynamics model to characterize the memory effects in system states. Building upon the Hegselmann-Krause framework and fractional-order difference, a comprehensive model is established that captures the persistent influence of historical information. Through rigorous theoretical analysis, the fundamental properties including convergence and consensus is investigated. The results demonstrate that the proposed model not only maintains favorable convergence and consensus characteristics compared to classical opinion dynamics, but also addresses limitations such as the monotonicity of bounded opinions. This enables a more realistic representation of opinion evolution in real-world scenarios. The findings of this study provide new insights and methodological approaches for understanding opinion formation and evolution, offering both theoretical significance and practical applications. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:23:06 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Jiang",
"Meiru",
""
],
[
"Su",
"Wei",
""
],
[
"Ren",
"Guojian",
""
],
[
"Yu",
"Yongguang",
""
]
] |
2506.04702 | Sakuntala Chatterjee | Shobhan Dev Mandal and Sakuntala Chatterjee | Bacterial Chemotaxis in a Traveling Wave Attractant Environment | null | J Stat Mech (2025) 063501 | 10.1088/1742-5468/add868 | null | q-bio.CB cond-mat.stat-mech physics.bio-ph | http://creativecommons.org/licenses/by/4.0/ | We study single cell E.coli chemotaxis in a spatio-temporally varying attractant environment. Modeling the attractant concentration in the form of a traveling sine wave, we measure in our simulations, the chemotactic drift velocity of the cell for different propagation speed of the attractant wave. We find a highly non-trivial dependence where the chemotactic drift velocity changes sign, and also shows multiple peaks. For slowly moving attractant wave, drift velocity is negative, i.e. the drift motion is directed opposite to wave propagation. As the wave speed increases, drift velocity shows a negative peak, then changes sign, reaches a positive peak and finally becomes zero when the wave moves too fast for the cell to respond. We explain this rich behavior from the difference in attractant gradient perceived by the cell during its run along the propagation direction and opposite to it. In particular, when the cell moves in the same direction as the wave, the relative velocity of the cell with respect to the wave becomes zero when the wave speed matches the run speed. In this limit, the cell is able to ride the wave and experiences no concentration gradient during these runs. On the contrary, for runs in the opposite direction, no such effect is present and the effective gradient increases monotonically with the wave speed. We show, using detailed quantitative measurements, how this difference gives rise to the counter-intuitive behavior of chemotactic drift velocity described above. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:25:51 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Mandal",
"Shobhan Dev",
""
],
[
"Chatterjee",
"Sakuntala",
""
]
] |
2506.04703 | Zening Yan | Zening Yan | Standardized Constraints on the Shadow Radius and the Instability of Scalar, Electromagnetic, $p$-Form, and Gravitational Perturbations of High-Dimensional Spherically Symmetric Black Holes in Einstein-power-Yang-Mills-Gauss-Bonnet Gravity | 70 pages, 22 figures, 4 tables | null | null | null | gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The space-time geometry under investigation is chosen to be a high-dimensional, static, spherically symmetric solution in an asymptotically flat background within the Einstein-power-Yang-Mills-Gauss-Bonnet (EPYMGB) gravity. To address the limitations of previous shadow constraints, we construct a standardized framework based on the Schwarzschild-Tangherlini metric to constrain the characteristic parameters of high-dimensional black holes by leveraging observational shadow data. Additionally, we provide a rigorous derivation of the shadow radius formula for a general high-dimensional spherically symmetric black hole. Subsequently, we systematically and comprehensively present the equations of motion and master variables governing spin-0, spin-1, $p$-form, and spin-2 perturbations in high-dimensional static spherically symmetric flat space-time. Our analysis reveals that the Yang-Mills magnetic charge $\mathcal{Q}$ and the power $q$ have a negligible impact on both the shadow radius and perturbations of the black hole when compared to the Gauss-Bonnet coupling constant ${\alpha}_2$ in various dimensions. Hence, the physical signatures of the parameters $\mathcal{Q}$ and $q$ in the black hole environment remain undetectable through either perturbation analysis or shadow observations. Cross-validation of the allowable range of ${\alpha}_2$ derived from the high-dimensional constraint on shadow radius and the dynamical stability analysis of gravitational perturbations demonstrates excellent agreement between these independent approaches. The conclusion of this cross-analysis further substantiate the accuracy of the high-dimensional shadow constraint formula proposed in this work, and we argue that this formula may serve as a universal tool for constraining the characteristic parameters of other high-dimensional spherically symmetric black hole solutions. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:26:06 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Yan",
"Zening",
""
]
] |
2506.04704 | Youngwan Lee | Youngwan Lee, Kangsan Kim, Kwanyong Park, Ilcahe Jung, Soojin Jang, Seanie Lee, Yong-Ju Lee, Sung Ju Hwang | HoliSafe: Holistic Safety Benchmarking and Modeling with Safety Meta Token for Vision-Language Model | Project page: https://youngwanlee.github.io/holisafe | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Despite emerging efforts to enhance the safety of Vision-Language Models (VLMs), current approaches face two main shortcomings. 1) Existing safety-tuning datasets and benchmarks only partially consider how image-text interactions can yield harmful content, often overlooking contextually unsafe outcomes from seemingly benign pairs. This narrow coverage leaves VLMs vulnerable to jailbreak attacks in unseen configurations. 2) Prior methods rely primarily on data-centric tuning, with limited architectural innovations to intrinsically strengthen safety. We address these gaps by introducing a holistic safety dataset and benchmark, HoliSafe, that spans all five safe/unsafe image-text combinations, providing a more robust basis for both training and evaluation. We further propose SafeLLaVA, a novel VLM augmented with a learnable safety meta token and a dedicated safety head. The meta token encodes harmful visual cues during training, intrinsically guiding the language model toward safer responses, while the safety head offers interpretable harmfulness classification aligned with refusal rationales. Experiments show that SafeLLaVA, trained on HoliSafe, achieves state-of-the-art safety performance across multiple VLM benchmarks. Additionally, the HoliSafe benchmark itself reveals critical vulnerabilities in existing models. We hope that HoliSafe and SafeLLaVA will spur further research into robust and interpretable VLM safety, expanding future avenues for multimodal alignment. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:26:34 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Lee",
"Youngwan",
""
],
[
"Kim",
"Kangsan",
""
],
[
"Park",
"Kwanyong",
""
],
[
"Jung",
"Ilcahe",
""
],
[
"Jang",
"Soojin",
""
],
[
"Lee",
"Seanie",
""
],
[
"Lee",
"Yong-Ju",
""
],
[
"Hwang",
"Sung Ju",
""
]
] |
2506.04705 | Junxian Lin | Junxian Lin, Zhao-Yu Li, Rui Guo, Jason A. S. Hunt, Teresa Antoja, Chengye Cao | Formation of the Two-Armed Phase Spiral from Multiple External Perturbations | 11 pages, 9 figures, Apj submitted, comments are welcomed | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent studies using the Gaia DR3 data have revealed a two-armed phase spiral in the $Z-V_Z$ phase space in the inner disk. In this study, we present new features of the two-armed phase spiral revealed by the Gaia Data and a new mechanism to explain such features with multiple external perturbations. By segmenting the Gaia DR3 RVS catalog based on $J_{\phi}$ (or $R_{g}$) and $\theta_\phi$, we confirm the existence of the clear two-armed phase spiral in the inner disk. Moreover, we identify a different two-armed phase spiral pattern at slightly larger radii, resembling a weak secondary branch along with the prominent major branch. At a given radius, with the azimuthal angle increasing, we observe a systematic transition of the two-armed phase spiral, with the significance of one branch weakened and another branch enhanced. This two-armed phase spiral may be due to the overlapping of distinct one-armed phase spirals. At different radii, the perturbation times estimated from each branch of the two-armed phase spiral are $\sim 320$ Myr and $\sim 500$ Myr, respectively, suggesting that the Galactic disk could be impacted by double external perturbers separated by $\sim 180$ Myr. We also performed test particle simulations of the disk perturbed by two satellite galaxies, which successfully generated a two-armed phase spiral similar to the observation. Both the observation and simulation results suggest that the signature in the $Z-V_Z$ phase space of earlier perturbations may not be completely erased by the more recent one. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:26:44 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Lin",
"Junxian",
""
],
[
"Li",
"Zhao-Yu",
""
],
[
"Guo",
"Rui",
""
],
[
"Hunt",
"Jason A. S.",
""
],
[
"Antoja",
"Teresa",
""
],
[
"Cao",
"Chengye",
""
]
] |
2506.04706 | Achyuta Rajaram | Achyuta Rajaram, Sarah Schwettmann, Jacob Andreas, and Arthur Conmy | Line of Sight: On Linear Representations in VLLMs | 8 pages, 9 figures | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Language models can be equipped with multimodal capabilities by fine-tuning on embeddings of visual inputs. But how do such multimodal models represent images in their hidden activations? We explore representations of image concepts within LlaVA-Next, a popular open-source VLLM. We find a diverse set of ImageNet classes represented via linearly decodable features in the residual stream. We show that the features are causal by performing targeted edits on the model output. In order to increase the diversity of the studied linear features, we train multimodal Sparse Autoencoders (SAEs), creating a highly interpretable dictionary of text and image features. We find that although model representations across modalities are quite disjoint, they become increasingly shared in deeper layers. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:30:58 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Rajaram",
"Achyuta",
""
],
[
"Schwettmann",
"Sarah",
""
],
[
"Andreas",
"Jacob",
""
],
[
"Conmy",
"Arthur",
""
]
] |
2506.04707 | Andreas H\"oring | Vladimiro Benedetti, Andreas H\"oring, Jie Liu | Intersection of two quadrics: modular interpretation and Hitchin morphism | 28 pages | null | null | null | math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The cotangent bundle $T^*X$ of a smooth intersection $X$ of two quadrics admits a Lagrangian fibration determined by the intrinsic geometry of $X$. We show that this fibration is actually the Hitchin morphism if we endow $X$ with a structure of moduli space of twisted Spin-bundles. This generalises the classical result for threefolds, in which case it recovers the Hitchin fibration for the moduli space of rank two bundles with fixed determinant of odd degree on a curve of genus two. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:31:06 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Benedetti",
"Vladimiro",
""
],
[
"Höring",
"Andreas",
""
],
[
"Liu",
"Jie",
""
]
] |
2506.04708 | Woomin Song | Woomin Song, Saket Dingliwal, Sai Muralidhar Jayanthi, Bhavana Ganesh, Jinwoo Shin, Aram Galstyan, Sravan Babu Bodapati | Accelerated Test-Time Scaling with Model-Free Speculative Sampling | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Language models have demonstrated remarkable capabilities in reasoning tasks through test-time scaling techniques like best-of-N sampling and tree search. However, these approaches often demand substantial computational resources, creating a critical trade-off between performance and efficiency. We introduce STAND (STochastic Adaptive N-gram Drafting), a novel model-free speculative decoding approach that leverages the inherent redundancy in reasoning trajectories to achieve significant acceleration without compromising accuracy. Our analysis reveals that reasoning paths frequently reuse similar reasoning patterns, enabling efficient model-free token prediction without requiring separate draft models. By introducing stochastic drafting and preserving probabilistic information through a memory-efficient logit-based N-gram module, combined with optimized Gumbel-Top-K sampling and data-driven tree construction, STAND significantly improves token acceptance rates. Extensive evaluations across multiple models and reasoning tasks (AIME-2024, GPQA-Diamond, and LiveCodeBench) demonstrate that STAND reduces inference latency by 60-65% compared to standard autoregressive decoding while maintaining accuracy. Furthermore, STAND outperforms state-of-the-art speculative decoding methods by 14-28% in throughput and shows strong performance even in single-trajectory scenarios, reducing inference latency by 48-58%. As a model-free approach, STAND can be applied to any existing language model without additional training, being a powerful plug-and-play solution for accelerating language model reasoning. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:31:18 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Song",
"Woomin",
""
],
[
"Dingliwal",
"Saket",
""
],
[
"Jayanthi",
"Sai Muralidhar",
""
],
[
"Ganesh",
"Bhavana",
""
],
[
"Shin",
"Jinwoo",
""
],
[
"Galstyan",
"Aram",
""
],
[
"Bodapati",
"Sravan Babu",
""
]
] |
2506.04709 | Jun-Jie Wei Dr. | Jun-Jie Wei, Xue-Feng Wu | Prospects for the Detection of High-Redshift Gamma-Ray Bursts in the Era of EP and SVOM | 10 pages, 5 figures, 3 tables | null | null | null | astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | Gamma-ray bursts (GRBs) are a promising probe of the high-redshift Universe, but their detection remains observationally challenging. In this work, we explore the detectability of high-$z$ GRBs by the Wide-field X-ray Telescope (WXT) aboard the Einstein Probe (\emph{EP}) and the coded-mask gamma-ray imager (ECLAIRs) aboard the Space-based multi-band astronomical Variable Objects Monitor (\emph{SVOM}). Using a population synthesis model calibrated to $Swift$ GRB observations, we develop a tool to estimate high-$z$ GRB detection rates for instruments with specific energy bands and sensitivities. Our results indicate that \emph{EP}/WXT could detect $\sim5.1^{+3.4}_{-2.4}$ (with 68\% confidence level) GRBs annually at $z>6$, compared to $\sim0.7^{+1.0}_{-0.4}$ $\mathrm{events\,yr^{-1}}$ at $z>6$ for \emph{SVOM}/ECLAIRs. While \emph{EP} cannot independently determine redshifts (requiring optical/near-infrared follow-up), its assumed $\sim30\%$ follow-up efficiency yields $\sim1.5^{+1.0}_{-0.7}$ confirmed $z>6$ GRBs annually. \emph{SVOM}, equipped with dedicated follow-up telescopes, will ensure robust high-$z$ GRB confirmations. We anticipate that \emph{EP} and \emph{SVOM} will open new avenues for utilizing enlarged samples of high-$z$ GRBs to explore the early Universe. Moreover, \emph{EP} will assemble a substantial sample of soft, low-luminosity GRBs at low-to-intermediate redshifts, providing critical insights into the structure of GRB jets. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:32:08 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Wei",
"Jun-Jie",
""
],
[
"Wu",
"Xue-Feng",
""
]
] |
2506.04710 | Lucas {\AA}kerstedt | Lucas {\AA}kerstedt, Harald Hultin, and B. L. G. Jonsson | An Array Decomposition Method for Finite Arrays with Electrically Connected Elements for fast Toeplitz Solvers | 12 pages, 17 figures | null | null | null | math.NA cs.NA eess.SP | http://creativecommons.org/licenses/by/4.0/ | A large part of the geometry of array antennas is often partially defined by finite translational symmetries. Applying the method of moments (MoM) with the RWG-like element on an appropriately structured mesh to these arrays results in an impedance matrix where the main part exhibits a multilevel block Toeplitz structure. This article introduces a memory-efficient construction method that effectively represents and reuses impedance calculations. The proposed method, applicable to electrically connected elements, also accounts for all non-symmetric parts of the array. The core idea involves nine distinct electrically connectable components from which the array can be assembled. The derived multilevel block Toeplitz matrix is further utilized by an in-house inverse solver to achieve faster and more memory-efficient MoM current vector calculations. We demonstrate the method by computing the far-field of a 32x32 array and the scattering parameters of two tightly coupled 9x9 arrays. This approach reduces the memory allocation from $\mathcal{O}(N_x^2 N_y^2)$ to $\mathcal{O}(N_x N_y)$, for an $N_x \times N_y$ array. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:34:17 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Åkerstedt",
"Lucas",
""
],
[
"Hultin",
"Harald",
""
],
[
"Jonsson",
"B. L. G.",
""
]
] |
2506.04711 | Te Ma | Te Ma, Min Bi, Saierdaer Yusuyin, Hao Huang and Zhijian Ou | LLM-based phoneme-to-grapheme for phoneme-based speech recognition | Interspeech 2025 | null | null | null | cs.SD cs.CL eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In automatic speech recognition (ASR), phoneme-based multilingual pre-training and crosslingual fine-tuning is attractive for its high data efficiency and competitive results compared to subword-based models. However, Weighted Finite State Transducer (WFST) based decoding is limited by its complex pipeline and inability to leverage large language models (LLMs). Therefore, we propose LLM-based phoneme-to-grapheme (LLM-P2G) decoding for phoneme-based ASR, consisting of speech-to-phoneme (S2P) and phoneme-to-grapheme (P2G). A challenge is that there seems to have information loss in cascading S2P and P2G. To address this challenge, we propose two training strategies: data augmentation with noisy phonemes (DANP), and randomized top-$K$ marginalized (TKM) training and decoding. Our experimental results show that LLM-P2G outperforms WFST-based systems in crosslingual ASR for Polish and German, by relative WER reductions of 3.6% and 6.9% respectively. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:35:55 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Ma",
"Te",
""
],
[
"Bi",
"Min",
""
],
[
"Yusuyin",
"Saierdaer",
""
],
[
"Huang",
"Hao",
""
],
[
"Ou",
"Zhijian",
""
]
] |
2506.04712 | Pinak Mandal | Pinak Mandal, Georg A. Gottwald | UNO: Unlearning via Orthogonalization in Generative models | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As generative models become increasingly powerful and pervasive, the ability to unlearn specific data, whether due to privacy concerns, legal requirements, or the correction of harmful content, has become increasingly important. Unlike in conventional training, where data are accumulated and knowledge is reinforced, unlearning aims to selectively remove the influence of particular data points without costly retraining from scratch. To be effective and reliable, such algorithms need to achieve (i) forgetting of the undesired data, (ii) preservation of the quality of the generation, (iii) preservation of the influence of the desired training data on the model parameters, and (iv) small number of training steps. We propose fast unlearning algorithms based on loss gradient orthogonalization. We show that our algorithms are able to forget data while maintaining the fidelity of the original model. Using MNIST and CelebA data, we demonstrate that our algorithms achieve orders of magnitude faster unlearning times than their predecessors, such as gradient surgery. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:37:02 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Mandal",
"Pinak",
""
],
[
"Gottwald",
"Georg A.",
""
]
] |
2506.04713 | Hanxin Wang | Hanxin Wang, Tian Liu, Shu Kong | Robust Few-Shot Vision-Language Model Adaptation | Project website: https://hannawang09.github.io/projects/srapf/ | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Pretrained VLMs achieve strong performance on downstream tasks when adapted with just a few labeled examples. As the adapted models inevitably encounter out-of-distribution (OOD) test data that deviates from the in-distribution (ID) task-specific training data, enhancing OOD generalization in few-shot adaptation is critically important. We study robust few-shot VLM adaptation, aiming to increase both ID and OOD accuracy. By comparing different adaptation methods (e.g., prompt tuning, linear probing, contrastive finetuning, and full finetuning), we uncover three key findings: (1) finetuning with proper hyperparameters significantly outperforms the popular VLM adaptation methods prompt tuning and linear probing; (2) visual encoder-only finetuning achieves better efficiency and accuracy than contrastively finetuning both visual and textual encoders; (3) finetuning the top layers of the visual encoder provides the best balance between ID and OOD accuracy. Building on these findings, we propose partial finetuning of the visual encoder empowered with two simple augmentation techniques: (1) retrieval augmentation which retrieves task-relevant data from the VLM's pretraining dataset to enhance adaptation, and (2) adversarial perturbation which promotes robustness during finetuning. Results show that the former/latter boosts OOD/ID accuracy while slightly sacrificing the ID/OOD accuracy. Yet, perhaps understandably, naively combining the two does not maintain their best OOD/ID accuracy. We address this dilemma with the developed SRAPF, Stage-wise Retrieval Augmentation-based Adversarial Partial Finetuning. SRAPF consists of two stages: (1) partial finetuning the visual encoder using both ID and retrieved data, and (2) adversarial partial finetuning with few-shot ID data. Extensive experiments demonstrate that SRAPF achieves the state-of-the-art ID and OOD accuracy on the ImageNet OOD benchmarks. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:37:15 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Wang",
"Hanxin",
""
],
[
"Liu",
"Tian",
""
],
[
"Kong",
"Shu",
""
]
] |
2506.04714 | Bhavana Akkiraju | Bhavana Akkiraju, Aishwarya Pothula, Santosh Kesiraju, Anil Kumar Vuppala | IIITH-BUT system for IWSLT 2025 low-resource Bhojpuri to Hindi speech translation | Paper is accepted to IWSLT2025 | null | null | null | cs.CL eess.AS | http://creativecommons.org/licenses/by/4.0/ | This paper presents the submission of IIITH-BUT to the IWSLT 2025 shared task on speech translation for the low-resource Bhojpuri-Hindi language pair. We explored the impact of hyperparameter optimisation and data augmentation techniques on the performance of the SeamlessM4T model fine-tuned for this specific task. We systematically investigated a range of hyperparameters including learning rate schedules, number of update steps, warm-up steps, label smoothing, and batch sizes; and report their effect on translation quality. To address data scarcity, we applied speed perturbation and SpecAugment and studied their effect on translation quality. We also examined the use of cross-lingual signal through joint training with Marathi and Bhojpuri speech data. Our experiments reveal that careful selection of hyperparameters and the application of simple yet effective augmentation techniques significantly improve performance in low-resource settings. We also analysed the translation hypotheses to understand various kinds of errors that impacted the translation quality in terms of BLEU. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:38:01 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Akkiraju",
"Bhavana",
""
],
[
"Pothula",
"Aishwarya",
""
],
[
"Kesiraju",
"Santosh",
""
],
[
"Vuppala",
"Anil Kumar",
""
]
] |
2506.04715 | Zelu Qi | Zelu Qi, Ping Shi, Chaoyang Zhang, Shuqi Wang, Fei Zhao, Da Pan, Zefeng Ying | Towards Holistic Visual Quality Assessment of AI-Generated Videos: A LLM-Based Multi-Dimensional Evaluation Model | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The development of AI-Generated Video (AIGV) technology has been remarkable in recent years, significantly transforming the paradigm of video content production. However, AIGVs still suffer from noticeable visual quality defects, such as noise, blurriness, frame jitter and low dynamic degree, which severely impact the user's viewing experience. Therefore, an effective automatic visual quality assessment is of great importance for AIGV content regulation and generative model improvement. In this work, we decompose the visual quality of AIGVs into three dimensions: technical quality, motion quality, and video semantics. For each dimension, we design corresponding encoder to achieve effective feature representation. Moreover, considering the outstanding performance of large language models (LLMs) in various vision and language tasks, we introduce a LLM as the quality regression module. To better enable the LLM to establish reasoning associations between multi-dimensional features and visual quality, we propose a specially designed multi-modal prompt engineering framework. Additionally, we incorporate LoRA fine-tuning technology during the training phase, allowing the LLM to better adapt to specific tasks. Our proposed method achieved \textbf{second place} in the NTIRE 2025 Quality Assessment of AI-Generated Content Challenge: Track 2 AI Generated video, demonstrating its effectiveness. Codes can be obtained at https://github.com/QiZelu/AIGVEval. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:40:12 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Qi",
"Zelu",
""
],
[
"Shi",
"Ping",
""
],
[
"Zhang",
"Chaoyang",
""
],
[
"Wang",
"Shuqi",
""
],
[
"Zhao",
"Fei",
""
],
[
"Pan",
"Da",
""
],
[
"Ying",
"Zefeng",
""
]
] |
2506.04716 | Hongyu Wang | Hongyu Wang, Yonghao Long, Yueyao Chen, Hon-Chi Yip, Markus Scheppach, Philip Wai-Yan Chiu, Yeung Yam, Helen Mei-Ling Meng, Qi Dou | Learning dissection trajectories from expert surgical videos via imitation learning with equivariant diffusion | null | null | 10.1016/j.media.2025.103599 | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Endoscopic Submucosal Dissection (ESD) is a well-established technique for removing epithelial lesions. Predicting dissection trajectories in ESD videos offers significant potential for enhancing surgical skill training and simplifying the learning process, yet this area remains underexplored. While imitation learning has shown promise in acquiring skills from expert demonstrations, challenges persist in handling uncertain future movements, learning geometric symmetries, and generalizing to diverse surgical scenarios. To address these, we introduce a novel approach: Implicit Diffusion Policy with Equivariant Representations for Imitation Learning (iDPOE). Our method models expert behavior through a joint state action distribution, capturing the stochastic nature of dissection trajectories and enabling robust visual representation learning across various endoscopic views. By incorporating a diffusion model into policy learning, iDPOE ensures efficient training and sampling, leading to more accurate predictions and better generalization. Additionally, we enhance the model's ability to generalize to geometric symmetries by embedding equivariance into the learning process. To address state mismatches, we develop a forward-process guided action inference strategy for conditional sampling. Using an ESD video dataset of nearly 2000 clips, experimental results show that our approach surpasses state-of-the-art methods, both explicit and implicit, in trajectory prediction. To the best of our knowledge, this is the first application of imitation learning to surgical skill development for dissection trajectory prediction. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:41:19 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Wang",
"Hongyu",
""
],
[
"Long",
"Yonghao",
""
],
[
"Chen",
"Yueyao",
""
],
[
"Yip",
"Hon-Chi",
""
],
[
"Scheppach",
"Markus",
""
],
[
"Chiu",
"Philip Wai-Yan",
""
],
[
"Yam",
"Yeung",
""
],
[
"Meng",
"Helen Mei-Ling",
""
],
[
"Dou",
"Qi",
""
]
] |
2506.04717 | Qiang Liu | Babar Hussain, Qiang Liu, Gang Chen, Bihai She, Dahai Yu | Using In-Context Learning for Automatic Defect Labelling of Display Manufacturing Data | null | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper presents an AI-assisted auto-labeling system for display panel defect detection that leverages in-context learning capabilities. We adopt and enhance the SegGPT architecture with several domain-specific training techniques and introduce a scribble-based annotation mechanism to streamline the labeling process. Our two-stage training approach, validated on industrial display panel datasets, demonstrates significant improvements over the baseline model, achieving an average IoU increase of 0.22 and a 14% improvement in recall across multiple product types, while maintaining approximately 60% auto-labeling coverage. Experimental results show that models trained on our auto-labeled data match the performance of those trained on human-labeled data, offering a practical solution for reducing manual annotation efforts in industrial inspection systems. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:42:31 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Hussain",
"Babar",
""
],
[
"Liu",
"Qiang",
""
],
[
"Chen",
"Gang",
""
],
[
"She",
"Bihai",
""
],
[
"Yu",
"Dahai",
""
]
] |
2506.04718 | Norimi Yokozaki | Xiao He, Takaaki Nomura and Norimi Yokozaki | Dark matter and dark radiation from chiral $U(1)$ gauge symmetry | 15 pages, 6 figures | null | null | null | hep-ph hep-ex | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider a simple model of a dark sector with a chiral $U(1)$ gauge symmetry. The anomaly-free condition requires at least five chiral fermions. Some of the fermions acquire masses through a vacuum expectation value of a Higgs field, and they are stable due to an accidental symmetry. This makes them dark matter candidates. If the dark sector was once in thermal equilibrium with the Standard Model and dark radiation constraints are included, two-component dark matter may be needed since the number of massless fermions is restricted. To meet the requirements from dark matter direct detection experiments, the main component should be a Majorana fermion, and the secondary component should be a Dirac fermion. The amount of Dirac fermion dark matter must be sufficiently small to satisfy direct detection limits. We also discuss the possibility of testing an invisible dark photon at future lepton collider experiments, taking into account cosmological constraints. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:43:51 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"He",
"Xiao",
""
],
[
"Nomura",
"Takaaki",
""
],
[
"Yokozaki",
"Norimi",
""
]
] |
2506.04719 | Farid Labib Dr. | Farid Labib, Hiroyuki Takakura, Asuka Ishikawa, Takenori Fujii and Ryuji Tamura | Derivation of a non-stoichiometric 1/1 quasicrystal approximant from a stoichiometric 2/1 quasicrystal approximant and maximization of magnetocaloric effect | 5 figures | null | null | null | cond-mat.mtrl-sci cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The present research introduces a novel strategy for tuning magnetic properties by overcoming the compositional limitation of stoichiometric intermetallic compounds via extension of their stability into a new dimension within valence electron-per-atom (e/a) parameter space. Focusing on approximant crystals (ACs), a "double hetero-valent elemental substitution" is employed in a stoichiometric Ga-Pt-Gd 2/1 AC whereby e/a is lowered from 1.92 to 1.60. Through this approach a new family of stable Ga-based Tsai-type 1/1 ACs with exceptionally wide composition stability within e/a space is derived. Remarkably, magnetic ground state is altered from initially spin-glass to ferromagnetic (FM) with second order phase transition and mean-field-like critical behavior. More importantly, through this strategy, the isothermal magnetic entropy change enhanced significantly and reached a maximum value of -8.7 J/K mol-Gd under a 5 T magnetic field change, even comparable to leading rare-earth magnetocaloric materials including RCo2 phases. These findings demonstrate the high potential of a double hetero-valent elemental substitution for tailoring magnetic properties and magnetocaloric response in stoichiometric compounds, offering a new pathway for designing high-performance magnetic refrigeration materials even beyond the quasicrystals and ACs. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:48:25 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Labib",
"Farid",
""
],
[
"Takakura",
"Hiroyuki",
""
],
[
"Ishikawa",
"Asuka",
""
],
[
"Fujii",
"Takenori",
""
],
[
"Tamura",
"Ryuji",
""
]
] |
2506.04720 | Anja Meyer | Anja Meyer | On the modular cohomology of $GL_2(\mathbb{Z}/p^n)$ and $SL_2(\mathbb{Z}/p^n)$ | null | null | null | null | math.AT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Let $p$ be an odd prime. Denote a Sylow $p$-subgroup of $GL_2(\mathbb{Z}/p^n)$ and $SL_2(\mathbb{Z}/p^n)$ by $S_p(n,GL)$ and $S_p(n,SL)$ respectively. The theory of stable elements tells us that the mod-$p$ cohomology of a finite group is given by the stable elements of the mod-$p$ cohomology of it's Sylow $p$-subgroup. We prove that for suitable group extensions of $S_p(n,GL)$ and $S_p(n,SL)$ the $E_2$-page of the Lyndon-Hochschild-Serre spectral sequence associated to these extensions does not depend on $n>1$. Finally, we use the theory of fusion systems to describe the ring of stable elements. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:50:05 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Meyer",
"Anja",
""
]
] |
2506.04721 | Yuru Jiang Miss. | Yuru Jiang, Wenxuan Ding, Shangbin Feng, Greg Durrett, Yulia Tsvetkov | SPARTA ALIGNMENT: Collectively Aligning Multiple Language Models through Combat | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | We propose SPARTA ALIGNMENT, an algorithm to collectively align multiple LLMs through competition and combat. To complement a single model's lack of diversity in generation and biases in evaluation, multiple LLMs form a "sparta tribe" to compete against each other in fulfilling instructions while serving as judges for the competition of others. For each iteration, one instruction and two models are selected for a duel, the other models evaluate the two responses, and their evaluation scores are aggregated through a adapted elo-ranking based reputation system, where winners/losers of combat gain/lose weight in evaluating others. The peer-evaluated combat results then become preference pairs where the winning response is preferred over the losing one, and all models learn from these preferences at the end of each iteration. SPARTA ALIGNMENT enables the self-evolution of multiple LLMs in an iterative and collective competition process. Extensive experiments demonstrate that SPARTA ALIGNMENT outperforms initial models and 4 self-alignment baselines across 10 out of 12 tasks and datasets with 7.0% average improvement. Further analysis reveals that SPARTA ALIGNMENT generalizes more effectively to unseen tasks and leverages the expertise diversity of participating models to produce more logical, direct and informative outputs. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:51:23 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Jiang",
"Yuru",
""
],
[
"Ding",
"Wenxuan",
""
],
[
"Feng",
"Shangbin",
""
],
[
"Durrett",
"Greg",
""
],
[
"Tsvetkov",
"Yulia",
""
]
] |
2506.04722 | Miyu Suzuki | Toshiki Matsusaka, Miyu Suzuki | Indefinite theta functions arising from affine Lie superalgebras and sums of triangular numbers | 39 pages | null | null | null | math.NT math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We extend the recently developed theory of Roehrig and Zwegers on indefinite theta functions to prove certain power series are modular forms. As a consequence, we obtain several power series identities for powers of the generating function of triangular numbers. We also show that these identities arise as specializations of denominator identities of affine Lie superalgebras. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:51:46 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Matsusaka",
"Toshiki",
""
],
[
"Suzuki",
"Miyu",
""
]
] |
2506.04723 | Jiayu Wang | Jiayu Wang, Yifei Ming, Zixuan Ke, Caiming Xiong, Shafiq Joty, Aws Albarghouthi, Frederic Sala | Beyond Accuracy: Dissecting Mathematical Reasoning for LLMs Under Reinforcement Learning | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Reinforcement learning (RL) has become the dominant paradigm for endowing language models with advanced reasoning capabilities. Despite the substantial empirical gains demonstrated by RL-based training methods like GRPO, a granular understanding of their advantages is still lacking. To address this gap, we introduce a fine-grained analytic framework to dissect the impact of RL on reasoning. Our framework specifically investigates key elements that have been hypothesized to benefit from RL training: (1) plan-following and execution, (2) problem decomposition, and (3) improved reasoning and knowledge utilization. Using this framework, we gain insights beyond mere accuracy. For instance, providing models with explicit step-by-step plans surprisingly degrades performance on the most challenging benchmarks, yet RL-tuned models exhibit greater robustness, experiencing markedly smaller performance drops than their base counterparts. This suggests that RL may not primarily enhance the execution of external plans but rather empower models to formulate and follow internal strategies better suited to their reasoning processes. Conversely, we observe that RL enhances the model's capacity to integrate provided knowledge into its reasoning process, leading to performance improvements across diverse tasks. We also study difficulty, showing improved training by developing new ways to exploit hard problems. Our findings lay a foundation for more principled training and evaluation of reasoning models. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:53:59 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Wang",
"Jiayu",
""
],
[
"Ming",
"Yifei",
""
],
[
"Ke",
"Zixuan",
""
],
[
"Xiong",
"Caiming",
""
],
[
"Joty",
"Shafiq",
""
],
[
"Albarghouthi",
"Aws",
""
],
[
"Sala",
"Frederic",
""
]
] |
2506.04724 | Giordano Scappucci | Davide Costa, Karina Hudson, Patrick Del Vecchio, Lucas E. A. Stehouwer, Alberto Tosato, Davide Degli Esposti, Mario Lodari, Stefano Bosco, Giordano Scappucci | Buried unstrained Ge channels: a lattice-matched platform for quantum technology | null | null | null | null | cond-mat.mes-hall | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Ge and Si strained quantum wells have enabled the most advanced spin-qubit quantum processors, but they are deposited on defective, metamorphic SiGe substrates, which may impact device performance and scaling. Here we introduce an alternative platform, based on a heterojunction between unstrained Ge and a strained SiGe barrier, which is lattice-matched to a Ge substrate. In a structure with a 52-nm-thick strained SiGe barrier, we demonstrate a low-disorder two-dimensional hole gas with a high-mobility of 1.33$\times$10$^5$ cm$^2$/Vs and a low percolation density of 1.4(1)$\times$10$^1$$^0$ cm$^-$$^2$. Quantum transport measurements show that confined holes have a strong density-dependent in-plane effective mass and out-of-plane $g$-factor, pointing to a significant heavy-hole--light-hole mixing in agreement with theory. The expected strong spin-orbit interaction, possibility of isotopic purification, and ability to host superconducting pairing correlations make this platform appealing for fast quantum hardware and hybrid quantum systems. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 07:58:30 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Costa",
"Davide",
""
],
[
"Hudson",
"Karina",
""
],
[
"Del Vecchio",
"Patrick",
""
],
[
"Stehouwer",
"Lucas E. A.",
""
],
[
"Tosato",
"Alberto",
""
],
[
"Esposti",
"Davide Degli",
""
],
[
"Lodari",
"Mario",
""
],
[
"Bosco",
"Stefano",
""
],
[
"Scappucci",
"Giordano",
""
]
] |
2506.04725 | Evgeny Perepelkin | E.E. Perepelkin, B.I. Sadovnikov, N.G. Inozemtseva, A.S. Medvedev | New exact solutions of the 3D Schr\"odinger equation | 32 pages, 6 figures | null | null | null | quant-ph math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | Previously we found a unique quantum system with a positive gauge-invariant Weyl-Stratonovich quasi-probability density function which can be defined by the so-called {\guillemotleft}quadratic funnel{\guillemotright} potential [Phys. Rev. A 110 02222 (2024)]. In this work we have constructed a class of exact solutions to the 3D Schr\"odinger equation for a two-parameter {\guillemotleft}quadratic funnel{\guillemotright} potential based on the -model of micro and macro systems. Explicit expressions for the energy spectrum and the set of eigenfunctions have been found. Using gauge invariance for scalar and vector potentials, a solution to the electromagnetic Schr\"odinger equation has been obtained, with a magnetic field in the form of a {\guillemotleft}Dirac string{\guillemotright} defined by a singular vortex probability flux field. Superpositions of eigenfunctions leading to various types of vortex and potential probability current fields have been investigated in detail. The analysis of the quantum system's properties has been carried out within the Wigner-Vlasov formalism. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:01:44 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Perepelkin",
"E. E.",
""
],
[
"Sadovnikov",
"B. I.",
""
],
[
"Inozemtseva",
"N. G.",
""
],
[
"Medvedev",
"A. S.",
""
]
] |
2506.04726 | Kiyoshi Kanazawa | Kiyoshi Kanazawa, Andreas Dechant | Stochastic thermodynamics for classical non-Markov jump processes | 5+3 pages, 4 figures | null | null | null | cond-mat.stat-mech | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Stochastic thermodynamics investigates energetic/entropic bounds in small systems, such as biomolecular motors, chemical-reaction networks, and quantum nano-devices. Foundational results, including the second law and thermodynamic uncertainty relations, predominantly rely on the Markov assumption -- neglecting history dependence of physical systems. However, while physicists recognise that the Markov assumption is dubious in real experimental setups, extending stochastic thermodynamics to general non-Markov systems has proven challenging due to their mathematical complexity. Here we establish the general theory of stochastic thermodynamics for arbitrary classical non-Markov jump processes. We introduce a key technique, called the {\it Fourier embedding}, which converts any non-Markov jump process into the Markov field dynamics of auxiliary Fourier modes. This approach yields the necessary and sufficient condition for time-reversal symmetry and enables the derivation of the second law for our non-Markov systems. Our framework accommodates diverse non-Markovian dynamics in realistic experimental setups and offers a guiding principle for physics-informed modelling of history-dependent fluctuations. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:02:17 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Kanazawa",
"Kiyoshi",
""
],
[
"Dechant",
"Andreas",
""
]
] |
2506.04727 | Souvik Ghosh | Souvik Ghosh | Numerical Investigation of Stub Length Influence on Dispersion Relations and Parity Effect in Aharonov-Bohm Rings | 8 pages, 3 figures | null | null | null | cond-mat.mtrl-sci cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | Aharonov-Bohm (AB) rings with side-attached stubs are model systems for quantum-interference studies in mesoscopic physics. The geometry of such systems, particularly the ratio of stub length ($v$) to ring circumference ($u$), can significantly alter their electronic states. In this work, we solve Deo's transcendental mode-condition equation (Eq. 2.15 from Deo, 2021 [Deo2021]) numerically -- using Python's NumPy and SciPy libraries -- for ring-stub geometries with $v/u = 0.200, 0.205,$ and $0.210$ to generate dispersion relations ($ku$ vs. $\Phi/\Phi_{0}$) and the underlying function $\text{Re}(1/T)$. We find that changing $v/u$ shifts several of the six lowest calculated dispersion branches, with $\Delta(ku)$ up to approximately $0.34$ for the 6th branch at $\Phi=0$ when comparing $v/u=0.200$ and $v/u=0.210$. This also alters gap widths. Notably, for $v/u=0.205$ and $v/u=0.210$, the 5th and 6th consecutive calculated modes both exhibit paramagnetic slopes near zero Aharonov-Bohm flux, indicating the parity breakdown initiates at or below $v/u=0.205$. This directly demonstrates a breakdown of the simple alternating parity effect predicted by Deo (2021) [Deo2021]. These results highlight the sensitivity of mesoscopic ring spectra to fine-tuning of stub length, with potential implications for experimental control of persistent currents, as further illustrated by calculations of the net current. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:02:34 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Ghosh",
"Souvik",
""
]
] |
2506.04728 | Xin Qian | Haobo Yang, Zhenguo Zhu, Zhongnan Xie, Jinhong Du, Shuo Bai, Hong Guo, Te-Huan Liu, Ronggui Yang and Xin Qian | Thermal Property Microscopy with Compressive Sensing Frequency-Domain Thermoreflectance | null | null | null | null | physics.app-ph cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Spatial mapping of thermal properties is critical for unveiling the structure-property relation of materials, heterogeneous interfaces, and devices. These property images can also serve as datasets for training artificial intelligence models for material discoveries and optimization. Here we introduce a high-throughput thermal property imaging method called compressive sensing frequency domain thermoreflectance (CS-FDTR), which can robustly profile thermal property distributions with micrometer resolutions while requiring only a random subset of pixels being experimentally measured. The high-resolution thermal property image is reconstructed from the raw down-sampled data through L_1-regularized minimization. The high-throughput imaging capability of CS-FDTR is validated using the following cases: (a) the thermal conductance of a patterned heterogeneous interface, (b) thermal conductivity variations of an annealed pyrolytic graphite sample, and (c) the sharp change in thermal conductivity across a vertical aluminum/graphite interface. With less than half of the pixels being experimentally sampled, the thermal property images measured using CS-FDTR show nice agreements with the ground truth (point-by-point scanning), with a relative deviation below 15%. This work opens the possibility of high-throughput thermal property imaging without sacrificing the data quality, which is critical for materials discovery and screening. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:03:32 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Yang",
"Haobo",
""
],
[
"Zhu",
"Zhenguo",
""
],
[
"Xie",
"Zhongnan",
""
],
[
"Du",
"Jinhong",
""
],
[
"Bai",
"Shuo",
""
],
[
"Guo",
"Hong",
""
],
[
"Liu",
"Te-Huan",
""
],
[
"Yang",
"Ronggui",
""
],
[
"Qian",
"Xin",
""
]
] |
2506.04729 | Sandro Mereghetti | Sandro Mereghetti (INAF, IASF-Milano) | Galactic Science -- Rapporteur Talk of the 8th Heidelberg International Symposium on High Energy Gamma Ray Astronomy | 8 pages, 2 tables. To appear in the Procs. of the 8th Heidelberg International Symposium on High-Energy Gamma-Ray Astronomy, held in Milan in 2024, to be published in Memorie della Societ\`a Astronomica Italiana (MemSAIt) | null | null | null | astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | The most recent observational and theoretical results in the rapidly expanding field of high-energy gamma-ray astrophysics were discussed at the international conference ``Gamma-2024'' that took place in Milano in September 2024. This contribution summarises the 'rapporteur talk' relative to the Galactic science given at the end of the conference. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:03:39 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Mereghetti",
"Sandro",
"",
"INAF, IASF-Milano"
]
] |
2506.04730 | Mohammadreza Azimi | M. R. Azimi, I. Akbarbaglu and A. R. Imanzadeh Fard | $J$-class weighted translations on locally compact groups | null | null | null | null | math.FA math.DS | http://creativecommons.org/licenses/by/4.0/ | A bounded linear operator $T$ on a Banach space $X$ (not necessarily separable) is said to be $J$-class operator whenever the extended limit set, say $J_T(x)$ equals $X$ for some vector $x\in X$. Practically, the extended limit sets localize the dynamical behavior of operators. In this paper, using the extended limit sets we will examine the necessary and sufficient conditions for the weighted translation $T_{a,\omega}$ to be $J$-class on a locally compact group $G$, within the setting of $ L^p$-spaces for $ 1 \leq p < \infty $. Precisely, we delineate the boundary between $J$-class and hypercyclic behavior for weighted translations. Then, we will show that for torsion elements in locally compact groups, unlike the case of non-dense orbits of weighted translations, we have $J_{T_{a,\omega}}(0)=L^p(G)$. Finally, we will provide some examples on which the weighted translation $ T_{a,\omega}$ is $J$-class but it fails to be hypercyclic. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:04:53 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Azimi",
"M. R.",
""
],
[
"Akbarbaglu",
"I.",
""
],
[
"Fard",
"A. R. Imanzadeh",
""
]
] |
2506.04731 | Elena V. Mikheeva | A. N. Melikhov (Astro Space Center, P. N. Lebedev Physical Institute, Russian Academy of Sciences, Moscow, Russia), E. V. Mikheeva (Astro Space Center, P. N. Lebedev Physical Institute, Russian Academy of Sciences, Moscow, Russia) | Constraints on the Number of Primordial Black Holes Due to Interaction with Dust | 5 pages, 3 figures | Astronomy Reports, Vol 66, No 5, pp 387-392 (2022) | 10.1134/S1063772922050043 | null | astro-ph.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Photons emitted by primordial black holes (PBHs) due to the Hawking effect are among the factors of interstellar dust heating. Based on the data on the temperature of dust, constraints on the fraction of $10^{15}\leq M\leq10^{17}$~g. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:05:57 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Melikhov",
"A. N.",
"",
"Astro Space Center, P. N. Lebedev Physical Institute, Russian Academy of Sciences, Moscow, Russia"
],
[
"Mikheeva",
"E. V.",
"",
"Astro Space Center, P. N. Lebedev Physical Institute, Russian Academy of Sciences, Moscow, Russia"
]
] |
2506.04732 | Nicola Cavallini Dr. | Nicola Cavallini, Gianmarco Manzini, Daniele Funaro, Andrea Favalli | A Fast, Accurate and Oscillation-free Spectral Collocation Solver for High-dimensional Transport Problems | null | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by/4.0/ | Transport phenomena-describing the movement of particles, energy, or other physical quantities-are fundamental in various scientific disciplines, including nuclear physics, plasma physics, astrophysics, engineering, and the natural sciences.
However, solving the associated seven-dimensional transport equations poses a significant computational challenge due to the curse of dimensionality.
We introduce the Tensor Train Superconsistent Spectral (T${^2}$S${^2}$) solver to address this challenge, integrating Spectral Collocation for exponential convergence, Superconsistency for stabilization in transport-dominated regimes, and Tensor Train format for substantial data compression. T${^2}$S${^2}$ enforces a dimension-wise superconsistent condition compatible with tensor structures, achieving extremely low compression ratios, in the order of $(10^{-12})$, while preserving spectral accuracy. Numerical experiments on linear problems demonstrate that T${^2}$S${^2}$ can solve high-dimensional transport problems in minutes on standard hardware, making previously intractable problems computationally feasible. This advancement opens new avenues for efficiently and accurately modeling complex transport phenomena. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:07:07 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Cavallini",
"Nicola",
""
],
[
"Manzini",
"Gianmarco",
""
],
[
"Funaro",
"Daniele",
""
],
[
"Favalli",
"Andrea",
""
]
] |
2506.04733 | Peter Braun-Munzinger | Peter Braun-Munzinger, Krzysztof Redlich, Johanna Stachel | The quark-gluon plasma: diagnosis with thermal hadron production from the early history until detailed characterization at high energy colliders | Contribution to: Quark Gluon Plasma at Fifty -- A commemorative Journey, Publisher: Springer Nature Switzerland AG, Editors: Tapan Nayak, Marco Van Leeuwen, Steffen Bass, Claudia Ratti, James Dunlop | null | null | null | nucl-th hep-ex hep-ph nucl-ex | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In nuclear collisions at relativistic energies, matter is created which resembles closely the matter that filled all space until about 15 microseconds after the big bang. Here we summarize selected aspects of the research that led to the establishment of this new sub-field of physics and briefly describe its current `state of the art' with emphasis on matter creation through thermal particle production. In particular, we will focus on particle production at low transverse momentum and explain how its analysis sheds light on one of the key open questions, i.e. what is the mechanism of hadronization of colored objects such as quarks and gluons. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:09:01 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Braun-Munzinger",
"Peter",
""
],
[
"Redlich",
"Krzysztof",
""
],
[
"Stachel",
"Johanna",
""
]
] |
2506.04734 | Yongfu Zhu | Lin Sun, Weihong Lin, Jinzhu Wu, Yongfu Zhu, Xiaoqi Jian, Guangxiang Zhao, Change Jia, Linglin Zhang, Sai-er Hu, Yuhan Wu, Xiangzheng Zhang | Evaluation is All You Need: Strategic Overclaiming of LLM Reasoning Capabilities Through Evaluation Design | null | null | null | null | cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | Reasoning models represented by the Deepseek-R1-Distill series have been widely adopted by the open-source community due to their strong performance in mathematics, science, programming, and other domains. However, our study reveals that their benchmark evaluation results are subject to significant fluctuations caused by various factors. Subtle differences in evaluation conditions can lead to substantial variations in results. Similar phenomena are observed in other open-source inference models fine-tuned based on the Deepseek-R1-Distill series, as well as in the QwQ-32B model, making their claimed performance improvements difficult to reproduce reliably. Therefore, we advocate for the establishment of a more rigorous paradigm for model performance evaluation and present our empirical assessments of the Deepseek-R1-Distill series models. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:09:11 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Sun",
"Lin",
""
],
[
"Lin",
"Weihong",
""
],
[
"Wu",
"Jinzhu",
""
],
[
"Zhu",
"Yongfu",
""
],
[
"Jian",
"Xiaoqi",
""
],
[
"Zhao",
"Guangxiang",
""
],
[
"Jia",
"Change",
""
],
[
"Zhang",
"Linglin",
""
],
[
"Hu",
"Sai-er",
""
],
[
"Wu",
"Yuhan",
""
],
[
"Zhang",
"Xiangzheng",
""
]
] |
2506.04735 | Wolf von Klitzing | Saurabh Pandey, Hector Mas, Georgios Vasilakis, and Wolf von Klitzingl | Atomtronic Matter-Wave Optics | 4 figures.5 pages, | Physical Review Letters 126 17 (2021) | 10.1103/physrevlett.126.170402 | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Matterwaves made up of ultra-cold quantum-degenerate atoms have enabled the creation of tools having unprecedented sensitivity and precision in measuring gravity, rotation or magnetic fields. Applications range from gravitational wave detection and tests of Einstein's equivalence principle to inertial sensing for navigation and gravitational gradient sensing for oil and mineral exploration. In this letter, we introduce atom-optics as a novel tool of manipulating matterwaves in ring-shaped coherent waveguides. We collimate and focus matterwaves derived from Bose-Einstein Condensates (BECs) and ultra-cold thermal atoms in ring-shaped time-averaged adiabatic potentials. We demonstrate `delta-kick cooling' of BECs, reducing their expansion energies by a factor of 34. The atomtronic waveguide ring has a radius of only $485\,\mu m$, compared to other state-of-the-art experiments requiring zero gravity or chambers of ten meter. This level of control with extremely reduced spatial requirements is an important step towards atomtronic quantum sensors. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:13:25 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Pandey",
"Saurabh",
""
],
[
"Mas",
"Hector",
""
],
[
"Vasilakis",
"Georgios",
""
],
[
"von Klitzingl",
"Wolf",
""
]
] |
2506.04736 | Xuan Ye | Xuan Ye, Yang Zhang, Bo Wang | Gravitational effects on Hong-Ou-Mandel interference in terrestrial laboratory | 31 pages, 12 figures | null | null | null | gr-qc | http://creativecommons.org/licenses/by/4.0/ | In this study, we investigate the relativistic effects of Earth on Hong-Ou-Mandel (HOM) interference experiments conducted in a terrestrial laboratory. Up to the second order, we calculate the relativistic time delay from the null geodesic equation (particle perspective), and the phase shift, along with the associated effective time delay, from the Klein-Gordon equation (wave perspective). Since gravity influences both the temporal and spatial parts of the phase shift, these time delays differ and predict different coincidence probabilities. The previous HOM experiment conducted on a rotating platform shows that the wave perspective can explain the experimental results. We further explore the frame-dragging and redshift effects within an arbitrarily oriented rectangular interferometer in two distinct cases, finding that both effects can be amplified by increasing the number of light loops. Additionally, we emphasize that the next-leading order Sagnac effect, due to the gravitational acceleration, is comparable to the Thomas precession, geodetic effect, and Lense-Thirring effect. To detect the leading-order Sagnac effect and the redshift effect caused by the gravitational acceleration within the current experimental precision, we determine the number of loops photons should travel in the interferometer. Furthermore, we suggest using the difference between any two HOM patterns with different effective time delays as a probe to detect the influence of relativistic effects on quantum systems. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:14:13 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Ye",
"Xuan",
""
],
[
"Zhang",
"Yang",
""
],
[
"Wang",
"Bo",
""
]
] |
2506.04737 | Mikhail Kennerley | Mikhail Kennerley, Angelica Alives-Reviro, Carola-Bibiane Sch\"onlieb, Robby T. Tan | Bridging Annotation Gaps: Transferring Labels to Align Object Detection Datasets | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Combining multiple object detection datasets offers a path to improved generalisation but is hindered by inconsistencies in class semantics and bounding box annotations. Some methods to address this assume shared label taxonomies and address only spatial inconsistencies; others require manual relabelling, or produce a unified label space, which may be unsuitable when a fixed target label space is required. We propose Label-Aligned Transfer (LAT), a label transfer framework that systematically projects annotations from diverse source datasets into the label space of a target dataset. LAT begins by training dataset-specific detectors to generate pseudo-labels, which are then combined with ground-truth annotations via a Privileged Proposal Generator (PPG) that replaces the region proposal network in two-stage detectors. To further refine region features, a Semantic Feature Fusion (SFF) module injects class-aware context and features from overlapping proposals using a confidence-weighted attention mechanism. This pipeline preserves dataset-specific annotation granularity while enabling many-to-one label space transfer across heterogeneous datasets, resulting in a semantically and spatially aligned representation suitable for training a downstream detector. LAT thus jointly addresses both class-level misalignments and bounding box inconsistencies without relying on shared label spaces or manual annotations. Across multiple benchmarks, LAT demonstrates consistent improvements in target-domain detection performance, achieving gains of up to +4.8AP over semi-supervised baselines. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:16:15 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Kennerley",
"Mikhail",
""
],
[
"Alives-Reviro",
"Angelica",
""
],
[
"Schönlieb",
"Carola-Bibiane",
""
],
[
"Tan",
"Robby T.",
""
]
] |
2506.04738 | Damianos Iosifidis | Damianos Iosifidis | On the role of the Parity Violating Hojman--Holst term in Gravity Theories | 7 pages, no figures | null | null | null | gr-qc hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study Parity Violating Gravity Theories whose gravitational Lagrangian is a generic function of the scalar curvature and the parity odd curvature pseudoscalar, commonly known as the Holst (or Hojmann) term. Generalizing some previous results in the literature, we explicitly show that if the Hessian of this function is non-degenerate, the initial non-Riemannian Theory is on-shell equivalent to a metric Scalar-Tensor Theory. The generic form of the kinetic coupling function and the scalar potential of the resulting Theory are explicitly found and reported. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:16:39 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Iosifidis",
"Damianos",
""
]
] |
2506.04739 | Ziyi Zhou | Ziyi Zhou, Xiaoming Zhang, Litian Zhang, Yibo Zhang, Zhenyu Guan, Chaozhuo Li, Philip S. Yu | Lifelong Evolution: Collaborative Learning between Large and Small Language Models for Continuous Emergent Fake News Detection | null | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The widespread dissemination of fake news on social media has significantly impacted society, resulting in serious consequences. Conventional deep learning methodologies employing small language models (SLMs) suffer from extensive supervised training requirements and difficulties adapting to evolving news environments due to data scarcity and distribution shifts. Large language models (LLMs), despite robust zero-shot capabilities, fall short in accurately detecting fake news owing to outdated knowledge and the absence of suitable demonstrations. In this paper, we propose a novel Continuous Collaborative Emergent Fake News Detection (C$^2$EFND) framework to address these challenges. The C$^2$EFND framework strategically leverages both LLMs' generalization power and SLMs' classification expertise via a multi-round collaborative learning framework. We further introduce a lifelong knowledge editing module based on a Mixture-of-Experts architecture to incrementally update LLMs and a replay-based continue learning method to ensure SLMs retain prior knowledge without retraining entirely. Extensive experiments on Pheme and Twitter16 datasets demonstrate that C$^2$EFND significantly outperforms existed methods, effectively improving detection accuracy and adaptability in continuous emergent fake news scenarios. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:17:55 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zhou",
"Ziyi",
""
],
[
"Zhang",
"Xiaoming",
""
],
[
"Zhang",
"Litian",
""
],
[
"Zhang",
"Yibo",
""
],
[
"Guan",
"Zhenyu",
""
],
[
"Li",
"Chaozhuo",
""
],
[
"Yu",
"Philip S.",
""
]
] |
2506.04740 | Taiki Haga | Taiki Haga | Time Glasses: Symmetry Broken Chaotic Phase with a Finite Gap | 27 pages, 20 figures | null | null | null | cond-mat.stat-mech quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce the time glass, a non-periodic analogue of the discrete time crystal that arises in periodically driven dissipative quantum many-body systems. This phase is defined by two key features: (i) spatial long-range order arising from the spontaneous breaking of an internal symmetry, and (ii) temporally chaotic oscillations of the order parameter, whose lifetime diverges with system size. To characterize the time glass phase, we focus on the spectral gap of the one-cycle (Floquet) Liouvillian, which determines the decay rate of the slowest relaxation mode. Numerical studies of periodically driven dissipative Ising models show that, in the time glass phase, the Liouvillian gap remains finite in the thermodynamic limit, in contrast to time crystals where the gap closes exponentially with system size. We further demonstrate that the Liouvillian gap converges to the decay rate of the order-parameter autocorrelation derived from the classical (mean-field) dynamics in the thermodynamic limit. This result establishes a direct correspondence between microscopic spectral features and emergent macroscopic dynamics in driven dissipative quantum systems. At first glance, the existence of a nonzero Liouvillian gap appears incompatible with the presence of indefinitely persistent chaotic oscillations. We resolve this apparent paradox by showing that the quantum R\'enyi divergence between a localized coherent initial state and the highly delocalized steady state grows unboundedly with system size. This divergence allows long-lived transients to persist even in the presence of a finite Liouvillian gap. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:18:17 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Haga",
"Taiki",
""
]
] |
2506.04741 | Leo Gross | Leonard-Alexander Lieske, Aaron H. Oechsle, Igor Ron\v{c}evi\'c, Ilias Gazizullin, Florian Albrecht, Matthias Krinninger, Leonhard Grill, Friedrich Esch, Leo Gross | Tip-induced nitrene generation | null | null | null | null | cond-mat.mtrl-sci cond-mat.mes-hall physics.chem-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We generated trinitreno-s-heptazine, a small molecule featuring three nitrene centers, by tip-induced chemistry from the precursor 2,5,8-triazido-s-heptazine on bilayer NaCl on Au(111). The precursor's azide groups were dissociated to form mono-, di- and trinitreno-s-heptazine, yielding molecules with one to three nitrene centers. The precursor and its products are characterized by atomic force microscopy and scanning tunnelling microscopy. Broken-symmetry DFT and configuration interaction calculations of inter- and intra-nitrene exchange couplings suggest a ferromagnetic coupling of the S = 1 nitrene centers, resulting in a high-spin septet ground state for neutral trinitreno-s-heptazine in the gas phase. On bilayer NaCl on Au(111), the combined results of experiments and theory suggest trinitreno-s-heptazine to be an anion with a sextet ground state. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:18:48 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Lieske",
"Leonard-Alexander",
""
],
[
"Oechsle",
"Aaron H.",
""
],
[
"Rončević",
"Igor",
""
],
[
"Gazizullin",
"Ilias",
""
],
[
"Albrecht",
"Florian",
""
],
[
"Krinninger",
"Matthias",
""
],
[
"Grill",
"Leonhard",
""
],
[
"Esch",
"Friedrich",
""
],
[
"Gross",
"Leo",
""
]
] |
2506.04742 | Oliver Lundqvist | Oliver G. S. Lundqvist and Fabricio Oliveira | Was Residual Penalty and Neural Operators All We Needed for Solving Optimal Control Problems? | null | null | null | null | math.OC cs.AI | http://creativecommons.org/licenses/by/4.0/ | Neural networks have been used to solve optimal control problems, typically by training neural networks using a combined loss function that considers data, differential equation residuals, and objective costs. We show that including cost functions in the training process is unnecessary, advocating for a simpler architecture and streamlined approach by decoupling the optimal control problem from the training process. Thus, our work shows that a simple neural operator architecture, such as DeepONet, coupled with an unconstrained optimization routine, can solve multiple optimal control problems with a single physics-informed training phase and a subsequent optimization phase. We achieve this by adding a penalty term based on the differential equation residual to the cost function and computing gradients with respect to the control using automatic differentiation through the trained neural operator within an iterative optimization routine. We showcase our method on nine distinct optimal control problems by training three separate DeepONet models, each corresponding to a different differential equation. For each model, we solve three problems with varying cost functions, demonstrating accurate and consistent performance across all cases. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:22:16 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Lundqvist",
"Oliver G. S.",
""
],
[
"Oliveira",
"Fabricio",
""
]
] |
2506.04743 | Shuhan Xu | Shuhan Xu, Siyuan Liang, Hongling Zheng, Yong Luo, Aishan Liu, Dacheng Tao | SRD: Reinforcement-Learned Semantic Perturbation for Backdoor Defense in VLMs | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Vision-Language Models (VLMs) have achieved remarkable performance in image captioning, but recent studies show they are vulnerable to backdoor attacks. Attackers can inject imperceptible perturbations-such as local pixel triggers or global semantic phrases-into the training data, causing the model to generate malicious, attacker-controlled captions for specific inputs. These attacks are hard to detect and defend due to their stealthiness and cross-modal nature. By analyzing attack samples, we identify two key vulnerabilities: (1) abnormal attention concentration on specific image regions, and (2) semantic drift and incoherence in generated captions. To counter this, we propose Semantic Reward Defense (SRD), a reinforcement learning framework that mitigates backdoor behavior without prior knowledge of triggers. SRD uses a Deep Q-Network to learn policies for applying discrete perturbations (e.g., occlusion, color masking) to sensitive image regions, aiming to disrupt the activation of malicious pathways. We design a semantic fidelity score as the reward signal, which jointly evaluates semantic consistency and linguistic fluency of the output, guiding the agent toward generating robust yet faithful captions. Experiments across mainstream VLMs and datasets show SRD reduces attack success rates to 5.6%, while preserving caption quality on clean inputs with less than 10% performance drop. SRD offers a trigger-agnostic, interpretable defense paradigm against stealthy backdoor threats in multimodal generative models. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:22:24 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Xu",
"Shuhan",
""
],
[
"Liang",
"Siyuan",
""
],
[
"Zheng",
"Hongling",
""
],
[
"Luo",
"Yong",
""
],
[
"Liu",
"Aishan",
""
],
[
"Tao",
"Dacheng",
""
]
] |
2506.04744 | Aykut Kayhan | Aykut Kayhan | Compact spacelike biconservative hypersurfaces in de Sitter space | null | null | null | null | math.DG | http://creativecommons.org/publicdomain/zero/1.0/ | In this paper, we investigate the geometry of compact spacelike biconservative hypersurfaces with constant scalar curvature in de Sitter space $\mathbb{S}_1^{m+1}(c)$, under some geometric constraints. Our results extend the understanding of rigidity properties of such hypersurfaces in pseudo-Riemannian settings. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:22:55 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Kayhan",
"Aykut",
""
]
] |
2506.04745 | Camilla Mannino | Camilla Mannino (NERV), Pierpaolo Sorrentino (INS), Mario Chavez (NERV), Marie-Costance Corsi (NERV) | Neuronal avalanches as a predictive biomarker of BCI performance: towards a tool to guide tailored training program | null | null | null | null | cs.HC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Brain-Computer Interfaces (BCIs) based on motor imagery (MI) hold promise for restoring control in individuals with motor impairments. However, up to 30% of users remain unable to effectively use BCIs-a phenomenon termed ''BCI inefficiency.'' This study addresses a major limitation in current BCI training protocols: the use of fixed-length training paradigms that ignore individual learning variability. We propose a novel approach that leverages neuronal avalanches-spatiotemporal cascades of brain activity-as biomarkers to characterize and predict user-specific learning mechanism. Using electroencephalography (EEG) data collected across four MI-BCI training sessions in 20 healthy participants, we extracted two features: avalanche length and activations. These features revealed significant training and taskcondition effects, particularly in later sessions. Crucially, changes in these features across sessions ($\Delta$avalanche length and $\Delta$activations) correlated significantly with BCI performance and enabled prediction of future BCI success via longitudinal Support Vector Regression and Classification models. Predictive accuracy reached up to 91%, with notable improvements after spatial filtering based on selected regions of interest. These findings demonstrate the utility of neuronal avalanche dynamics as robust biomarkers for BCI training, supporting the development of personalized protocols aimed at mitigating BCI illiteracy. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:23:02 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Mannino",
"Camilla",
"",
"NERV"
],
[
"Sorrentino",
"Pierpaolo",
"",
"INS"
],
[
"Chavez",
"Mario",
"",
"NERV"
],
[
"Corsi",
"Marie-Costance",
"",
"NERV"
]
] |
2506.04746 | Wang Baiqiao | Fei Ding, Baiqiao Wang, Zijian Zeng, Youwei Wang | Multi-Layer GRPO: Enhancing Reasoning and Self-Correction in Large Language Models | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Group Relative Policy Optimization (GRPO) algorithm has demonstrated considerable success in enhancing the reasoning capabilities of large language models (LLMs), as evidenced by DeepSeek-R1. However, the absence of intermediate supervision in GRPO frequently leads to inefficient exploration dynamics. A single error in a complex reasoning chain can invalidate the entire solution, resulting in abrupt reward vanishing and compromising training stability.To address these challenges, we propose MGRPO (Multi-layer GRPO). MGRPO operates in two layers: the first layer employs standard GRPO to generate an initial response. This response, along with the original query, is then fed into a second-layer GRPO process. This second layer is specifically trained to identify and correct errors in the initial response, effectively creating a self-correction loop. This mechanism provides implicit process-level supervision by rewarding successful error correction, without requiring an explicit, densely-annotated reward model. Experimental results on several mathematical reasoning benchmarks demonstrate that MGRPO significantly outperforms standard GRPO, achieving superior performance by fostering both reasoning and self-correction abilities. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:27:34 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Ding",
"Fei",
""
],
[
"Wang",
"Baiqiao",
""
],
[
"Zeng",
"Zijian",
""
],
[
"Wang",
"Youwei",
""
]
] |
2506.04747 | Alexander Hedge | A. J. Hedge, N. Seymour, J. W. Broderick, A. Gupta, J. Afonso, L. Ighina, M. Lehnert, G. Noirot, S. Shabala, D. Stern, R. J. Turner | The Centre of Attention: a Powerful Radio Galaxy Pinpoints a NIR-Dark Protocluster at z~3.9 | null | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | We report the discovery of a $z\sim3.9$ protocluster identified from Atacama Large Millimetre/sub-millimetre Array Band 3 spectral scans of a bright radio source selected from the GaLactic and Extra-galactic All-sky Murchison Widefield Array (GLEAM) survey. Extended CO(4-3) and [CI](1-0) line emission was detected in GLEAM J005332$-$325630 confirming it to be a $z=3.879$ powerful radio galaxy with luminosity, $L_{500 MHz}=1.3\times10^{28}$ ${W Hz}^{-1}$. This source is part of a sample of candidate high redshift radio galaxies with bright radio fluxes, $S_{150MHz}>0.1$ Jy, but host galaxies with $K_s({AB})\gtrsim23$ mag. The molecular gas associated with the radio galaxy host has two kinematically separate components, likely in-falling and indicative of a recent interaction or merger with another galaxy. One 100-GHz continuum source $\sim120$ pkpc away is found to have both CO(4-3) and [CI](1-0) emission lines and a further five protocluster members are identified from CO(4-3) emission alone, all at similar redshift ($\Delta v<700$ km s$^{-1}$) and within a radius of $1.1^{\prime}$. Using photometry from the High Acuity Widefield K-band Imager $K_s$-band and the Dark Energy Survey $g, r, i, z$ and $Y$ bands, we find this protocluster harbours a rare, optically-dark, very massive $M_*\sim10^{12}$ ${M}_\odot$ galaxy. Comparisons with the TNG300 cosmological simulation puts this galaxy in a dark matter halo of $M_{DM}\sim3\times10^{13}$ ${M}_\odot$ which will evolve into a Coma-like DM halo ($M_{DM}\sim10^{15}$ ${M}_\odot$) by the present day. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:31:28 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Hedge",
"A. J.",
""
],
[
"Seymour",
"N.",
""
],
[
"Broderick",
"J. W.",
""
],
[
"Gupta",
"A.",
""
],
[
"Afonso",
"J.",
""
],
[
"Ighina",
"L.",
""
],
[
"Lehnert",
"M.",
""
],
[
"Noirot",
"G.",
""
],
[
"Shabala",
"S.",
""
],
[
"Stern",
"D.",
""
],
[
"Turner",
"R. J.",
""
]
] |
2506.04748 | Nanami Kotani | Nanami Kotani, Kuniharu Sakurada, Jiayi Xu, Masahiko Inami and Yasuaki Monnai | Synchro-Thermography: Monitoring ~10 mK Facial Temperature Changes with Heartbeat Referencing for Physiological Sensing | This paper has been submitted to the 2025 SICE Festival with Annual Conference (SICE FES 2025) | null | null | null | physics.med-ph eess.SP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Infrared thermography has gained interest as a tool for non-contact measurement of blood circulation and skin blood flow due to cardiac activity. Partiularly, blood vessels on the surface, such as on the back of the hand, are suited for visualization. However, standardized methodologies have not yet been established for areas such as the face and neck, where many blood vessels are lie deeper beneath the surface, and external stimulation for measurement could be harmful. Here we propose Synchro-Thermography for stable monitoring of facial temperature changes associated with heart rate variability. We conducted experiments with eight subjects and measured minute temperature changes with an amplitude of about \SI{10}{mK} on the forehead and chin. The proposed method improves the temperature resolution by a factor of 2 or more, and can stably measure skin temperature changes caused by blood flow. This skin temperature change could be applied to physiological sensing such as blood flow changes due to injury or disease, or as an indicator of stress. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:31:41 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Kotani",
"Nanami",
""
],
[
"Sakurada",
"Kuniharu",
""
],
[
"Xu",
"Jiayi",
""
],
[
"Inami",
"Masahiko",
""
],
[
"Monnai",
"Yasuaki",
""
]
] |
2506.04749 | Laurence Davies | Laurence Davies, Dan Mackinlay, Rafael Oliveira, Scott A. Sisson | Amortized variational transdimensional inference | 34 pages, 11 figures | null | null | null | stat.CO stat.ME stat.ML | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The expressiveness of flow-based models combined with stochastic variational inference (SVI) has, in recent years, expanded the application of optimization-based Bayesian inference to include problems with complex data relationships. However, until now, SVI using flow-based models has been limited to problems of fixed dimension. We introduce CoSMIC, normalizing flows (COntextually-Specified Masking for Identity-mapped Components), an extension to neural autoregressive conditional normalizing flow architectures that enables using a single amortized variational density for inference over a transdimensional target distribution. We propose a combined stochastic variational transdimensional inference (VTI) approach to training CoSMIC flows using techniques from Bayesian optimization and Monte Carlo gradient estimation. Numerical experiments demonstrate the performance of VTI on challenging problems that scale to high-cardinality model spaces. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:33:07 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Davies",
"Laurence",
""
],
[
"Mackinlay",
"Dan",
""
],
[
"Oliveira",
"Rafael",
""
],
[
"Sisson",
"Scott A.",
""
]
] |
2506.04750 | Ajay Soni | Aditya Singh, Divya Rawat, Victor Hjort, Abhisek Mishra, Arnaud le Febvrier, Subhankar Bedanta, Per Eklund, and Ajay Soni | Lattice Mismatch Driven In Plane Strain Engineering for Enhanced Upper Critical Fields in Mo2N Superconducting Thin Films | null | null | null | null | cond-mat.supr-con cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Transition metal nitrides are a fascinating class of hard coating material that provide an excellent platform for investigating superconductivity and fundamental electron phonon interactions. In this work the structural morphological and superconducting properties have been studied for Mo2N thin films deposited via direct current magnetron sputtering on cplane Al2O3 and MgO substrates to elucidate the effect of internal strain on superconducting properties. High resolution X Ray diffraction and time of flight elastic recoil detection analysis confirms the growth of single phase Mo2N thin films exhibiting epitaxial growth with twin domain structure. Low temperature electrical transport measurements reveal superconducting transitions at 5.2 K and 5.6 K with corresponding upper critical fields of 5 T and 7 T for the films deposited on Al2O3 and MgO, respectively. These results indicate strong type II superconductivity and the observed differences in superconducting properties are attributed to substrate induced strain which leads to higher e ph coupling for the film on MgO substrate. These findings highlight the tunability of superconducting properties in Mo2N films through strategic substrate selection. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:34:19 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Singh",
"Aditya",
""
],
[
"Rawat",
"Divya",
""
],
[
"Hjort",
"Victor",
""
],
[
"Mishra",
"Abhisek",
""
],
[
"Febvrier",
"Arnaud le",
""
],
[
"Bedanta",
"Subhankar",
""
],
[
"Eklund",
"Per",
""
],
[
"Soni",
"Ajay",
""
]
] |
2506.04751 | Ar Rohim | Ar Rohim, Apriadi Salim Adam | Fermionic Casimir effect in the presence of compact dimensions in field theory with Lorentz invariant violation | 22 pages, 18 Figures | null | null | null | hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this study, we investigate the effect of the Lorentz invariant violation on the Casimir energy and pressure of the massive fermion field in the presence of the compact dimensions with topological $R^4\times S^1$, referring to the Kaluza-Klein model. In the system, the Dirac field is confined between two parallel plates with the geometry described by MIT bag boundary conditions, and the compactified dimension satisfies quasi-periodic boundary conditions. We investigate two directions of the Lorentz violation, namely, space- and time-like. The results reveal that in the space-like vector case, the Lorentz violation's strength and the extra dimension affect the Casimir energy and pressure. In contrast, in the time-like vector case, they are only affected by the extra dimension. We also propose an indirect method to estimate the size of the extra dimension by comparing the frequency shift of the massless fermionic case to that of the scaled experimental data for the electromagnetic field. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:37:11 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Rohim",
"Ar",
""
],
[
"Adam",
"Apriadi Salim",
""
]
] |
2506.04752 | Shenghai Yuan | Tianxin Hu, Xinhang Xu, Thien-Minh Nguyen, Fen Liu, Shenghai Yuan, and Lihua Xie | Tire Wear Aware Trajectory Tracking Control for Multi-axle Swerve-drive Autonomous Mobile Robots | Accepted in Journal of Automation and Intelligence | null | null | null | cs.RO math.OC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Multi-axle Swerve-drive Autonomous Mobile Robots (MS-AGVs) equipped with independently steerable wheels are commonly used for high-payload transportation. In this work, we present a novel model predictive control (MPC) method for MS-AGV trajectory tracking that takes tire wear minimization consideration in the objective function. To speed up the problem-solving process, we propose a hierarchical controller design and simplify the dynamic model by integrating the \textit{magic formula tire model} and \textit{simplified tire wear model}. In the experiment, the proposed method can be solved by simulated annealing in real-time on a normal personal computer and by incorporating tire wear into the objective function, tire wear is reduced by 19.19\% while maintaining the tracking accuracy in curve-tracking experiments. In the more challenging scene: the desired trajectory is offset by 60 degrees from the vehicle's heading, the reduction in tire wear increased to 65.20\% compared to the kinematic model without considering the tire wear optimization. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:38:52 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Hu",
"Tianxin",
""
],
[
"Xu",
"Xinhang",
""
],
[
"Nguyen",
"Thien-Minh",
""
],
[
"Liu",
"Fen",
""
],
[
"Yuan",
"Shenghai",
""
],
[
"Xie",
"Lihua",
""
]
] |
2506.04753 | Niki Martinel | Niki Martinel and Rita Pucci | Physics Informed Capsule Enhanced Variational AutoEncoder for Underwater Image Enhancement | null | null | null | null | cs.CV cs.AI eess.IV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We present a novel dual-stream architecture that achieves state-of-the-art underwater image enhancement by explicitly integrating the Jaffe-McGlamery physical model with capsule clustering-based feature representation learning. Our method simultaneously estimates transmission maps and spatially-varying background light through a dedicated physics estimator while extracting entity-level features via capsule clustering in a parallel stream. This physics-guided approach enables parameter-free enhancement that respects underwater formation constraints while preserving semantic structures and fine-grained details. Our approach also features a novel optimization objective ensuring both physical adherence and perceptual quality across multiple spatial frequencies. To validate our approach, we conducted extensive experiments across six challenging benchmarks. Results demonstrate consistent improvements of $+0.5$dB PSNR over the best existing methods while requiring only one-third of their computational complexity (FLOPs), or alternatively, more than $+1$dB PSNR improvement when compared to methods with similar computational budgets. Code and data \textit{will} be available at https://github.com/iN1k1/. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:39:17 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Martinel",
"Niki",
""
],
[
"Pucci",
"Rita",
""
]
] |
2506.04754 | Mohd Tahseen | Mohd Tahseen, Vivekanand Dabade | Revisiting cofactor conditions: Elimination of transition layers in compound domains | null | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | This paper investigates the conditions necessary for the elimination of transition layers at interfaces involving compound domains, extending the classical framework of cofactor conditions. Although cofactor conditions enable stress-free phase boundaries between Type I/II domains and austenite, their applicability to compound domains has remained limited. Here, we present a comprehensive theoretical framework to characterize all compatible interfaces, highlighting the fundamental importance of the commutation property among martensitic variants. By establishing necessary and sufficient algebraic conditions, referred to as extreme compatibility conditions, we demonstrate the simultaneous elimination of transition layers at phase interfaces for both Type I/II and compound laminates, across all volume fractions of the martensitic variants. We also investigate the possibility of achieving supercompatibility in non-conventional twins, recently observed in the NiMnGa system. The focus of our work is on cubic-to-orthorhombic and cubic-to-monoclinic~II phase transformations, for which the extreme compatibility conditions are explicitly derived and systematically analyzed. The theory predicts novel zero-elastic-energy microstructures, including an increased number of triple clusters, spearhead-shaped martensitic nuclei, stress-free inclusions of austenite within martensite, and distinctive four-fold martensitic clusters. This significantly expands the possible modes of forming stress-free interfaces between phases and reveals new energy-minimizing microstructures that can facilitate the nucleation of martensite within austenite and vice versa. These configurations highlight significant enhancements in transformation reversibility and material durability, guiding the rational design of next-generation shape memory alloys with optimized functional properties. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:40:09 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Tahseen",
"Mohd",
""
],
[
"Dabade",
"Vivekanand",
""
]
] |
2506.04755 | Shenshen Li | Shenshen Li, Kaiyuan Deng, Lei Wang, Hao Yang, Chong Peng, Peng Yan, Fumin Shen, Heng Tao Shen, Xing Xu | Truth in the Few: High-Value Data Selection for Efficient Multi-Modal Reasoning | null | null | null | null | cs.CV cs.AI cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While multi-modal large language models (MLLMs) have made significant progress in complex reasoning tasks via reinforcement learning, it is commonly believed that extensive training data is necessary for improving multi-modal reasoning ability, inevitably leading to data redundancy and substantial computational costs. However, can smaller high-value datasets match or outperform full corpora for multi-modal reasoning in MLLMs? In this work, we challenge this assumption through a key observation: meaningful multi-modal reasoning is triggered by only a sparse subset of training samples, termed cognitive samples, whereas the majority contribute marginally. Building on this insight, we propose a novel data selection paradigm termed Reasoning Activation Potential (RAP), which identifies cognitive samples by estimating each sample's potential to stimulate genuine multi-modal reasoning by two complementary estimators: 1) Causal Discrepancy Estimator (CDE) based on the potential outcome model principle, eliminates samples that overly rely on language priors by comparing outputs between multi-modal and text-only inputs; 2) Attention Confidence Estimator (ACE), which exploits token-level self-attention to discard samples dominated by irrelevant but over-emphasized tokens in intermediate reasoning stages. Moreover, we introduce a Difficulty-aware Replacement Module (DRM) to substitute trivial instances with cognitively challenging ones, thereby ensuring complexity for robust multi-modal reasoning. Experiments on six datasets show that our RAP method consistently achieves superior performance using only 9.3% of the training data, while reducing computational costs by over 43%. Our code is available at https://github.com/Leo-ssl/RAP. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:40:24 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Li",
"Shenshen",
""
],
[
"Deng",
"Kaiyuan",
""
],
[
"Wang",
"Lei",
""
],
[
"Yang",
"Hao",
""
],
[
"Peng",
"Chong",
""
],
[
"Yan",
"Peng",
""
],
[
"Shen",
"Fumin",
""
],
[
"Shen",
"Heng Tao",
""
],
[
"Xu",
"Xing",
""
]
] |
2506.04756 | Ngoc Ly | Loan Dao, Ngoc Quoc Ly | Ontology-based knowledge representation for bone disease diagnosis: a foundation for safe and sustainable medical artificial intelligence systems | null | null | null | null | cs.AI cs.CV eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Medical artificial intelligence (AI) systems frequently lack systematic domain expertise integration, potentially compromising diagnostic reliability. This study presents an ontology-based framework for bone disease diagnosis, developed in collaboration with Ho Chi Minh City Hospital for Traumatology and Orthopedics. The framework introduces three theoretical contributions: (1) a hierarchical neural network architecture guided by bone disease ontology for segmentation-classification tasks, incorporating Visual Language Models (VLMs) through prompts, (2) an ontology-enhanced Visual Question Answering (VQA) system for clinical reasoning, and (3) a multimodal deep learning model that integrates imaging, clinical, and laboratory data through ontological relationships. The methodology maintains clinical interpretability through systematic knowledge digitization, standardized medical terminology mapping, and modular architecture design. The framework demonstrates potential for extension beyond bone diseases through its standardized structure and reusable components. While theoretical foundations are established, experimental validation remains pending due to current dataset and computational resource limitations. Future work will focus on expanding the clinical dataset and conducting comprehensive system validation. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:41:23 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Dao",
"Loan",
""
],
[
"Ly",
"Ngoc Quoc",
""
]
] |
2506.04757 | Riccardo Seppi | R. Seppi, D. Eckert, A. Finoguenov, S .Shreeram, E. Tempel, G. Gozaliasl, M. Lorenz, J. Wilms, G. A. Mamon, F. Gastaldello, L. Lovisari, E. O'Sullivan, K. Kolokythas, M. A. Bourne, M. Sun, and A. Pillepich | Modelling the selection of galaxy groups with end to end simulations | Accepted for publication on A&A | null | null | null | astro-ph.CO astro-ph.GA astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Feedback from supernovae and AGN shapes galaxy formation and evolution, yet its impact remains unclear. Galaxy groups offer a crucial probe, as their binding energy is comparable to that available from their central AGN. The XMM-Newton Group AGN Project (X-GAP) is a sample of 49 groups selected in X-ray (ROSAT) and optical (SDSS) bands, providing a benchmark for hydrodynamical simulations. In sight of such a comparison, understanding selection effects is essential. We aim to model the selection function of X-GAP by forward modelling the detection process in the X-ray and optical bands. Using the Uchuu simulation, we build a halo light cone, predict X-ray group properties with a neural network trained on hydro simulations, and assign galaxies matching observed properties. We compare the selected sample to the parent population. Our method provides a sample that matches the observed distribution of X-ray luminosity and velocity dispersion. The 50% completeness is reached at a velocity dispersion of 450 km/s in the X-GAP redshift range. The selection is driven by X-ray flux, with secondary dependence on velocity dispersion and redshift. We estimate a 93% purity level in the X-GAP parent sample. We calibrate the velocity dispersion-halo mass relation. We find a normalisation and slope in agreement with the literature, and an intrinsic scatter of about 0.06 dex. The measured velocity dispersion is accurate within 10% only for rich systems with more than about 20 members, while the velocity dispersion for groups with less than 10 members is biased at more than 20%. The X-ray follow-up refines the optical selection, enhancing purity but reducing completeness. In an SDSS-like setup, velocity dispersion measurement errors dominate over intrinsic scatter. Our selection model will enable the comparisons of thermodynamic properties and gas fractions between X-GAP groups and hydro simulations. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:42:33 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Seppi",
"R.",
""
],
[
"Eckert",
"D.",
""
],
[
"Finoguenov",
"A.",
""
],
[
"Shreeram",
"S .",
""
],
[
"Tempel",
"E.",
""
],
[
"Gozaliasl",
"G.",
""
],
[
"Lorenz",
"M.",
""
],
[
"Wilms",
"J.",
""
],
[
"Mamon",
"G. A.",
""
],
[
"Gastaldello",
"F.",
""
],
[
"Lovisari",
"L.",
""
],
[
"O'Sullivan",
"E.",
""
],
[
"Kolokythas",
"K.",
""
],
[
"Bourne",
"M. A.",
""
],
[
"Sun",
"M.",
""
],
[
"Pillepich",
"A.",
""
]
] |
2506.04758 | Yijun Cao | Yijun Cao, Fuya Luo, and Yongjie Li | Toward Better SSIM Loss for Unsupervised Monocular Depth Estimation | 12 pages,4 figures | International Conference on Image and Graphics. Cham: Springer Nature Switzerland, 2023: 81-92 | 10.1007/978-3-031-46305-1_7 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Unsupervised monocular depth learning generally relies on the photometric relation among temporally adjacent images. Most of previous works use both mean absolute error (MAE) and structure similarity index measure (SSIM) with conventional form as training loss. However, they ignore the effect of different components in the SSIM function and the corresponding hyperparameters on the training. To address these issues, this work proposes a new form of SSIM. Compared with original SSIM function, the proposed new form uses addition rather than multiplication to combine the luminance, contrast, and structural similarity related components in SSIM. The loss function constructed with this scheme helps result in smoother gradients and achieve higher performance on unsupervised depth estimation. We conduct extensive experiments to determine the relatively optimal combination of parameters for our new SSIM. Based on the popular MonoDepth approach, the optimized SSIM loss function can remarkably outperform the baseline on the KITTI-2015 outdoor dataset. | [
{
"version": "v1",
"created": "Thu, 05 Jun 2025 08:43:24 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Cao",
"Yijun",
""
],
[
"Luo",
"Fuya",
""
],
[
"Li",
"Yongjie",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.