Peer Revenue Play credits. Real papers.

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Wednesday 9 September

Computation and Language

Towards Stress-Aware Sentence-Level Filipino G2P With Weakly-Supervised ByT5 Fine-Tuning

Grapheme-to-phoneme conversion (G2P) refers to the task of converting a sequence of graphemes to a corresponding sequence of phonemes. While Filipino G2P is fairly straightforward due to its shallow orthography, the inclusion of prosodic features such as stress adds a layer of complexity that requires sentence-level context instead of single-word inputs. However, sentence-level data for Filipino typically do not include phoneme transcriptions, posing a challenge for training G2P models. As such, we investigate how to obtain sentence-level phoneme data for Filipino using available data and compare the resulting models with multilingual word-level G2P as well as measure how accurately they predict stress marker position for Filipino. We propose fine-tuning a ByT5-based model, pre-trained on multilingual word-level G2P data, on three sentence-level G2P datasets annotated with an LLM-assisted pipeline guided by data from Wiktionary. This approach produces models that perform well on the G2P task, achieving at best around 0.54% PER and 2.50% CER, a significant decrease compared to base model PER at around 19.74%, on a manually-corrected test set. The model is able to correctly classify most of the main stress classes in Filipino, but struggles particularly with malumi words. We show that a ByT5-based model performs well at sentence-level Filipino G2P and offers strong potential for Filipino homograph disambiguation.
Read the full abstract

All authors Lorenz Bernard Marqueses, Paulo Grane Gabriel Silva, Chastine Cabatay, Ericson Adler Tan, Ann Franchesca Laguna

Computer Vision

Putting Captions to the Test: Evaluating Video Caption Quality through Multiple-Choice Question Answering

Evaluating video captioning remains a critical challenge for Visual Large Language Models (VLLMs). Existing metrics primarily rely on matching generated text against ground-truth references. This paradigm suffers from the “one-to-many” nature of video description, where high-quality captions are often penalized for lexical mismatches or valid shifts in visual focus. Furthermore, such assessments are typically one-dimensional, failing to provide a fine-grained analysis of caption quality. To address this, we redefine caption quality through the lens of information fidelity: A caption must maximize the coverage of salient visual information while ensuring strict factuality. We introduce CapQuiz, a novel reference-free benchmark that assesses captions based on their utility in answering human-verified, fine-grained, multiple-choice questions derived from the video. CapQuiz features a hierarchical taxonomy of 10 question types (spanning Descriptive and Inferential categories) across 24 diverse video domains. Extensive experiments demonstrate that CapQuiz correlates significantly better with human judgments than existing metrics and offers interpretable insights into model performance.
Read the full abstract

All authors Zizhen Wang, Bo Feng, Zhengfeng Lai, Shiyu Li, Yang Lu, Meng Cao, Ping Huang, Xiaoming Simon Wang

Probability

Percolation of the contact process on the regular tree

The contact process on the regular tree $\mathbb{T}_d$ when $d\geq 3$ has the two phase transitions of global and of local survival, found by Pemantle and Liggett at values $\lambda_1$ and $\lambda_2$. We start the system with every vertex infected and let it relax to what is known as the upper invariant infection. In this stationary state, $\lambda_p$ is the critical value beyond which the infected vertices can percolate through $\mathbb{T}_d$, and $\lambda_{p^\complement}$ is the parameter before which the healthy vertices can percolate. We find these are both distinct phase transitions \[0<\lambda_1<\lambda_p<\lambda_2<\lambda_{p^\complement}<+\infty\] on $\mathbb{T}_d$ when $d\geq 7$. The most interesting of these comparisons is $\lambda_1<\lambda_p$, which we find for all $d\geq 3$. This comparison $\lambda_1<\lambda_p$ is a long-standing open question on $\mathbb{Z}^d$ with $d\geq 2$ and was not yet found on any other graphs except where $\lambda_p$ is infinite.
Read the full abstract

Machine Learning

Deep Neural Networks for Learning Intent from sEMG Signals to Support Hardware Devices for Post-Stroke Neurorehabilitation

Finger-specific motor intent is a clinically meaningful control signal for post-stroke neurorehabilitation, where residual muscle activity may remain measurable despite weak or incomplete movement. We study five-finger multilabel intent decoding from impaired-arm high-density surface electromyography (sEMG) in PhysioMio, a bilateral longitudinal dataset collected from stroke patients. A common processing protocol aligns movement labels, applies 20–450 Hz Butterworth filtering and Symlet-4 wavelet denoising, segments overlapping 200 ms windows, and extracts twelve time- and frequency-domain descriptors per channel. Direct LSTM, CNN, and GNN baselines reveal complementary behavior: the LSTM attains the highest subset accuracy (0.545), whereas the GNN attains the highest macro F1 (0.706) and macro AUPRC (0.776). Architecture search then identifies CNN-Large as the strongest single-split CNN, with 0.593 subset accuracy and 0.714 macro F1, while CNN-Micro provides a compact architecture for embedded inference. To match a four-sensor hardware design, we retrain CNN-Micro using channels associated with ECRB, ECRL, FDS, and FDP and exclude the ground electrode from model input. Across five seeds, cross-channel knowledge distillation improves the four-channel student over direct training, reaching $0.5219 \pm 0.0114$ subset accuracy, $0.7612 \pm 0.0038$ finger accuracy, and $0.6095 \pm 0.0058$ macro F1. The selected 123K-parameter model accepts nine windows of 48 features and has been exported to ONNX. These results establish a reproducible software path from post-stroke sEMG to compact five-finger intent prediction for subsequent hardware-in-the-loop evaluation.
Read the full abstract

All authors Zakariyya Brewster, Divy Wadhwani, Emily Yan, Aidan Wang, Karma Namgyal, Shuting Xie, Markiyan Konyk, Tala Abdelmaguid

Earth and Planetary Astrophysics

Can the Long-Term Impact of Stellar M-Dwarf Flares Alter the Spectral Features of a Giant Gaseous Exoplanet?

In this work, we model the long-term impact of recurrent stellar flares on the atmospheres of metal-rich gaseous exoplanets. Using synthetic flare spectra from a fiducial flare model integrated with a photochemical kinetics code, we track the changes in atmospheric composition with time. We further analyze the spectral variability by feeding these abundance profiles into a radiative transfer code at various time steps. Our simulations showed variability and persistent changes in key atmospheric species, such as CH4, CO2, and SO2, when compared to their quiescent state. Extreme flare events cause rapid depletion of molecules in the upper atmosphere and a temporary disappearance of spectral features, especially the SO2 feature at 7-8 microns, which shifted by about 75 ppm. Many species did not fully return to their quiescent state after flares, resulting in lasting changes in abundance, especially for SO2 and CO2, key species when inferring the atmospheric metallicty. We also explored the cumulative effects of recurrent flares, showing that species like H2O and CH4 followed a decreasing abundance trend, with half-lives of around 28 to 31 years. These results indicate that flare activity plays a significant role in shaping both the short- and long-term atmospheric composition and spectral features of giant gaseous exoplanets orbiting M-dwarf stars, underscoring the need to account for stellar activity when characterising such atmospheres. Our findings also highlight that the atmosphere of the modelled planet is not static, suggesting that a probabilistic approach to atmospheric abundances may be more appropriate than static retrievals, particularly for gaseous planets orbiting active stars.
Read the full abstract

Information Theory

The kernel-block rank profile and a complete classification of $\mathbb{Z}_2\mathbb{Z}_4\mathbb{Z}_8$-linear Hadamard codes

The $\mathbb{Z}_2\mathbb{Z}_4\mathbb{Z}_8$-additive codes are subgroups of $\mathbb{Z}_2^{\alpha_1}\times\mathbb{Z}_4^{\alpha_2}\times\mathbb{Z}_8^{\alpha_3}$, and a $\mathbb{Z}_2\mathbb{Z}_4\mathbb{Z}_8$-linear Hadamard code is the Gray map image of such a code. A recursive construction of $\mathbb{Z}_2\mathbb{Z}_4\mathbb{Z}_8$-additive Hadamard codes $\mathcal H^{t_1,t_2,t_3}$, with all $\alpha_i\neq0$, $t_1\geq1$, $t_2\geq0$, and $t_3\geq1$, is known, as are the linearity, kernel dimension, and rank of the corresponding codes $H^{t_1,t_2,t_3}$ of length $2^t$, where $t+1=3t_1+2t_2+t_3$. Yet these invariants do not completely classify the family. Two infinite families of pairs of distinct types share the length, rank and kernel dimension, and for classified lengths $2^t$, $3\leq t\leq11$, such pairs were separated only by computer equivalence tests. In this paper, we introduce an equivalence invariant that resolves these cases. The kernel partitions the binary coordinates into blocks, two coordinates lying in the same block when every kernel word takes the same value in both; the kernel-block rank profile is the multiset of the dimensions of the linear span punctured on these blocks. Unlike rank and kernel dimension, which are global, this invariant records how much of the span survives on each block. We compute it for the whole family: there are $2^{t_1+t_2+t_3-1}$ blocks, all of size $2^{2t_1+t_2}$, and the profile takes at most two values $t_2+\binom{t_1+2}{2}$ and $t_2+2+\binom{t_1+1}{2}$, whose difference is $t_1-1$; it is constant precisely when $t_1=1$. Hence, the profile recovers $t_1$, and the length and kernel dimension recover $t_2$ and $t_3$. Two codes of the family with the same length are therefore equivalent if and only if their types coincide, and the number of pairwise nonequivalent such codes of length $2^t$ is $\lfloor(t^2+6)/12\rfloor$ for every $t\geq3$.
Read the full abstract

Group Theory

The spectrum of conjugator length functions

A recent program tries to find which functions appear as conjugator length functions. In this note, we show that any (computable) increasing function larger than $n$ appears as $\mathrm{Cl}_G$ for some finitely generated (recursively presented) group. On the other hand, we demonstrate that either $\mathrm{Cl}_G$ must be constant or $\mathrm{Cl}_G(n)\succ n$. Combining these, we obtain a complete description of which functions appear as conjugator length functions of finitely generated groups.
Read the full abstract

Analysis of PDEs

Classification of multi-solitons in one sense of time for the 5D energy-critical wave equation

We classify the multi-solitons, in the positive sense of time, of the focusing energy-critical wave equation in space dimension $5$, more precisely the solutions $u(t,x)$ of the equation \[ \partial_t^2 u - \Delta u - |u|^{\frac 4{3}} u = 0, \quad (t,x)\in [T_0,\infty)\times {\mathbb R}^5, \] which satisfy the estimate \[ \left\|\nabla_{t,x} \left(u(t) - \sum_{k} W_k(t) \right) \right\|_{L^2} \lesssim t^{-\frac12^+} \quad \hbox{for all $t\gg 1$}. \] Here, $\{W_k\}_k$ is any given finite family of traveling waves, based on the explicit ground state solution \[ W(x) = \left(1+ \frac{|x|^2}{15}\right)^{-\frac32} \] with collinear two-by-two different velocities.
Read the full abstract

Analysis of PDEs

The $C^{p'}$-regularity conjecture near $p=2$

We prove the $C^{p'}$-regularity conjecture in every dimension when $p>2$ is sufficiently close to $2$. To this end, we establish improved Hölder estimates for the gradients of $p$-harmonic functions. These estimates also determine the first-order asymptotics of the optimal gradient Hölder exponent in every dimension. The proof combines compactness, harmonic rigidity of the limiting profiles, and a sharp uniform gap estimate for the first variation of the gradient excess.
Read the full abstract

Audio and Speech Processing

Over-Tightening-Aware Pseudo-Labeling for Tight-Boundary Speaker Diarization

Training speaker diarization models on loose labels, such as speech segments with padded boundaries or filled pauses, often results in similarly loose model outputs. To obtain tighter boundaries, pseudo-labeling based on the averaged outputs of causal and anticausal models has been proposed. However, since the pseudo-labels are estimation-based, they can suffer from over-tightening, which increases missed detections that can propagate as unrecoverable errors to downstream tasks. This paper carefully analyzes the causes of over-tightening and proposes three approaches to address them: (i) removing pause filling rather than padding, (ii) introducing a burn-in phase to mitigate missed detections near the beginning of causal and anticausal predictions, and (iii) making pseudo-label-based co-training aware of the non-causal model used for final inference. Experimental results show that the proposed method reduces missed detections caused by over-tightening and improves both diarization accuracy and downstream multi-talker ASR performance.
Read the full abstract

All authors Shota Horiguchi, Takanori Ashihara, Marc Delcroix, Naohiro Tawara, Alexis Plaquet

Computation and Language

5-Dialects-BN: Unmasking the Impact of Transliteration on Bangla Dialectal LLMs

Large Language Models (LLMs) have achieved remarkable progress across natural language processing (NLP) tasks, yet their capabilities degrade sharply for low-resource languages and dialectally diverse settings. Bangla, the world's sixth most spoken language, exemplifies this gap: existing resources overwhelmingly target Standard Bangla, leaving its regional dialects without the benchmarks needed to develop or evaluate dialect-aware systems. We address this gap with 5-Dialects-BN, the first multi-annotation Bangla dialect benchmark to align Romanized transliteration with dialectal text, Standard Bangla, English, and subjectivity labels across five regional varieties. The dataset comprises 6,000 manually annotated entries spanning five major dialects: Chittagong, Barisal, Noakhali, Sylhet, and Rangpur (Chittagong 1,900; Noakhali 1,500; Sylhet 1,200; Barisal 700; Rangpur 700), reflecting natural online availability. Each entry is enriched with five aligned annotations: the original dialectal text, a Romanized transliteration, an English translation, a Standard Bangla translation, and a subjectivity label (subjective vs. objective). Annotations were produced and cross-validated by native speakers and undergraduate linguistics students to ensure dialectal authenticity and semantic fidelity. The resulting resource supports a diverse suite of tasks, including dialect identification, dialect-to-standard normalization, machine translation, subjectivity classification, and parameter-efficient fine-tuning (e.g., LoRA) of multilingual LLMs. By providing a standardized, multi-annotation benchmark, 5-Dialects-BN enables principled evaluation of LLMs on dialectally diverse Bangla and lays a foundation for further research in low-resource, dialect-aware NLP.
Read the full abstract

All authors Md Mahir Jawad, Galib Mahmud Jim, Rafid Ahmed, Mir Sazzat Hossain, Md Fahim, Md Farhad Alam Bhuiyan

Cryptography and Security

CrossLink: Breaking Location Privacy by Linking Device Identifiers Across Protocols

Smartphones simultaneously transmit temporary identifiers over LTE, WiFi, and BLE. Existing privacy defenses analyze identifier randomization per protocol, implicitly assuming that these protections compose across protocols. We show that they do not: Even when each protocol leaks only temporary identifiers and the adversary is fully passive, unsynchronized identifier rotations allow cross-protocol stitching of device traces. We present CrossLink, an uncertainty-aware tracing algorithm that links identifiers across time, space, and protocols under noisy localization and mobility. We evaluate CrossLink using controlled lab experiments with commodity devices and large-scale mobility simulation. Under large-scale mobility simulation, CrossLink reconstructs full traces for 83% of users, versus 22% for the best single-protocol baseline, showing that location privacy must be analyzed jointly across protocols. We further show that CrossLink remains effective under partial coverage: strategically placed sniffers near LTE handover regions, mobile sniffers, and limited high-coverage subregions retain sufficient cross-protocol evidence to bridge observation gaps, achieving substantially higher linkability than random deployments.
Read the full abstract

All authors Aneet Kumar Dutta, Mihirraj Dixit, Kevin Gni, Wouter Lueks, Mridula Singh

Instrumentation and Detectors

A unified quantum electrical platform for synchronous metrological realization of volt, ohm and ampere

A co-located integration quantum electrical standard is essential to reduce reliance on distributed traceability in high-accuracy metrology, especially for portable and on-site use. Metrologically, realizing any two of voltage, resistance, and current is sufficient, as the third follows from Ohm's law. The combination of Josephson voltage and quantum Hall resistance offers better uncertainty, but conflicts with the tesla-level field for quantum Hall and near-zero field for Josephson operation. Here we report a compact unified platform enabling co-realization of quantum voltage and resistance in a single cryostat near 4 K, with quantum current derived via Ohm's law. A hierarchical magnetic shielding with staged attenuation and spatial confinement allows 6 T and below 50 nT to coexist within 270 mm axial separation with negligible cross-coupling. In integrated operation, the Josephson and quantum Hall subsystems agree with expected quantized values within relative standard uncertainties of 2.6E-9 and 1.4E-8, respectively. Linking them via an improved cryogenic current comparator realizes a 50 {\mu}A quantum current with relative uncertainty of 6.6E-8. These results demonstrate that three basic electrical units can be synchronously realized with superior metrological consistency on a single integrated platform, offering a viable transition from distributed calibration chains toward compact-integrated quantum-based realization.
Read the full abstract

All authors Lei Wang, Lushuai Qian, Yunfeng Lu, Yang Shi, Xuanlu Yang, Xiaoding Huang, Zihan Lei, Yuan Zhang, Yaqiong Fu, Junsheng Cheng, Jianhua Liu, Xinning Hu, Yinming Dai, Jianting Zhao, Qiuliang Wang

Cryptography and Security

Decentralized network congestion control for DAG-based distributed ledger system

We propose a variable and behavior-based node-specific proof-of-work (PoW) model for a directed acyclic graph (DAG)-based distributed ledger technology (DLT) network to mitigate decentralized network congestion control. Network congestion control for centralized communication systems is an established field of study, with detailed and continuous research being done on the subject. However, attention to congestion control in decentralized networks is relatively recent and underexplored, especially with DLT, such as blockchain and DAG-based networks. For the DLT networks, the network congestion is caused by factors such as transaction spamming, an increase in the user base, and the launch of new tokens. We focus on the congestion caused by the spamming of transactions within the blockchain and DAG-based DLT network. Based on the network throughput of transactions per second and consensus procedure, the DAG-based DLT needs to control network spamming more than the blockchain network. The PoW model within the DLT consensus framework is a limited deterrent against spamming. Our model provides equal opportunities for all stakeholders regardless of their computational resources. It prevents and penalizes any node that attempts to spam or dominate the network with more than the prescribed number of transactions. Since the system nodes compete to issue transactions with finite network resources, we display the system behavior through a non-cooperative game. Further, we show that our model enforces prescribed behavior amongst the nodes through the proof of the existence of Nash equilibrium in the game.
Read the full abstract

All authors Mayank Pandey, Rachit Agarwal, Sandeep Kumar Shukla, Nishchal Kumar Verma

Human-Computer Interaction

Somatosensory Activation and Attentional States in Creative Making

The methods for capturing the creative process come with associated tensions around memory recall, articulation, and communication during the act of making, as well as how to record these considerations. This paper has a twofold purpose: first, to offer an example of a mixed methodology, drawn from dance anthropology, sensory ethnography, and design, that applies embodied methods as an alternative for documenting creative making. Specifically, this incorporates the researcher-as-participant and the collation of fieldnotes, embodied knowledge/movement recall, with notation forms, and participant interviews. These are existing methods in dance anthropology; however, using them alongside exploratory prototyping and workshop approaches broadened this work into transdisciplinary practice. Second, it discusses the activation of somatosensory systems through wearable technology and the facilitation of heightened sensory awareness for the practitioner, leading to a subsequent ability to focus on creative decisions linked to reflection and metacognition.
Read the full abstract

Optimization and Control

Null-controllability for some linear parabolic systems with controls acting on different parts of the domain and its boundary

In this work we study the null-controllability properties of linear parabolic systems with constant coefficients in the case where several controls are acting on different distributed subdomains and/or on the boundary. We prove a Kalman rank condition in the one-dimensional case. In the case where only distributed controls are considered we also establish related results such as a Carleman estimate.
Read the full abstract

Chemical Physics

More is not always better: Dissociative photoionization limits the EUV absorbing photoacid generator pentafluorophenyl triflate in photolithography

Pentafluorophenyl triflate has been explored as a highly absorbing neutral photoacid generator (PAG) candidate for next generation chemically amplified resists used in extreme ultraviolet (EUV) lithography. Although increased fluorination enhances EUV absorption, this study demonstrates that such an approach does not necessarily improve photoacid generation efficiency. Using photoelectron-photoion coincidence (PEPICO) spectroscopy at the 92 eV photon energy of the EUV scanners in combination with quantum chemical calculations, the dissociative photoionization of pentafluorophenyl triflate was systematically investigated. The photoionization mass spectrum reveals extensive fragmentation, with the parent ion contributing only 3.1 % of the total signal and CF$_3^+$ representing the dominant product ion. Computed appearance energies align well with experimental trends and support a sequential fragmentation pathway involving loss of SO$_2$, CF$_3$, and CO. Crucially, none of the major dissociation channels yield precursors capable of forming triflic acid, the strong photoacid required for efficient deprotection reactions in chemically amplified resists. Combined with previous dissociative electron attachment studies indicating similarly unfavorable fragmentation, the results demonstrate that despite its high EUV absorption cross section, pentafluorophenyl triflate is unsuitable as a PAG for EUV lithography. The findings highlight the importance of understanding fundamental photoionization and electron interaction mechanisms to guide the rational design of next generation high performance EUV photoresists.
Read the full abstract

All authors Fabian Holzmeier, Michiel J. van Setten, Ragnar Björnsson, Robert Richter, Filipe Ferreira da Silva, Oddur Ingólfsson

Optimization and Control

Beyond Verified Answers: Solver-Informed Self-Distillation for Bootstrapping Operations Research Language Models

Modern large language models (LLMs) can translate natural-language descriptions into operations research (OR) formulations. Post-training techniques including reinforcement learning and on-policy self-distillation have further improved this capability. However, three limitations remain in training LLMs for OR formulations. First, training commonly relies on synthetic formulations validated by human experts or stronger models, constraining scalable supervision. Second, credit assignment is either coarse or costly: outcome rewards score an entire trajectory without locating the responsible modeling decision, whereas process-level supervision requires an additional evaluator. Third, privileged self-distillation can induce style mismatch by using solver context unavailable at deployment. We find that a model can improve from solver-artifact feedback generated by its own rollouts, making self-distillation a practical, evaluator-free source of dense supervision. Therefore, we propose SOLID: Solver-Informed On-Policy LearnIng through Self-Distillation, a novel framework for self-improving OR language models without verified answers or external evaluators. SOLID executes candidate programs from multiple rollouts, clusters their objectives, and selects a majority-group artifact as a pseudo-reference. The model then performs updates using group-relative advantages and dense self-supervision signals. Across multiple OR benchmarks, SOLID improves solution accuracy for both general-purpose and OR-tuned models over outcome-only group-relative training. These results show that solver artifacts can support scalable self-improvement without trusted answers.
Read the full abstract

All authors Rui Zhu, Minglong Cao, Chenyu Zhou, Jianghao Lin, Dongdong Ge

Materials Science

Strong Impact of Halide Ordering on Structural Phase Transitions in Mixed Perovskites

Mixed halide perovskites are highly versatile semiconductors with applications in photovoltaics, light-emitting diodes, and photodetectors. Understanding their thermodynamic phase behavior is central to guiding compositional design and improving device stability. Here, we train machine-learned interatomic potentials (MLIPs) on density functional theory reference data for CsxRb1-xPbBr3yI3-3y, CsxRb1-xPbBr3yCl3-3y, and CsxRb1-xPbCl3yI3-3y halide perovskites, enabling large-scale hybrid Monte Carlo-molecular dynamics simulations that sample both configurational and vibrational degrees of freedom. All three binary halide systems exhibit a miscibility gap, the extent of which correlates with halide ion size mismatch. The gaps in Br-Cl and Br-I close at low temperatures, while the Cl-I gap extends above room temperature. At temperatures above the miscibility gap (200 K to 500 K), all systems show a tendency toward layered halide ordering, with halide species preferentially occupying apical or equatorial octahedral sites. In CsPbBr3yI3-3y, this ordering occurs in a device-relevant temperature regime and is linked to the structural phase transitions, shifting transition temperatures by up to 100 K relative to randomly mixed structures. We attribute the strongly non-linear composition dependence of the orthorhombic-tetragonal phase boundary observed experimentally (a linear decrease followed by a plateau) to halide ordering. Introducing Rb on the A-site weakens halide ordering and eliminates the non-linear behavior, while narrowing the miscibility gap in both the Br-I and Br-Cl systems. These results establish halide ordering as a key determinant of structural phase stability in mixed-halide perovskites.
Read the full abstract

All authors Felix Uddén, Erik Fransson, Julia Wiktor, Benjamin M. Gallant, Dominik J. Kubicki, Paul Erhart

Astrophysics of Galaxies

Milky Way Structure from Double White Dwarf Gravitational-Wave Sources

The millihertz gravitational-wave sky will be dominated by $\sim10^{8}$ double white dwarfs while $\sim10^{4}$ be resolvable in the Milky Way, whose three-dimensional spatial distribution traces the Galaxy's structural parameters. We present Galena, a hierarchical Bayesian pipeline that recovers these parameters from a double white dwarf catalogue through an inhomogeneous Poisson-process likelihood. Each source's $3\times3$ Galactic position covariance is computed from the waveform Fisher information matrix and propagated analytically through a $3\times5$ Jacobian; the resulting measurement-error convolution is factored into a pre-computed sparse weight matrix, rendering the nested-sampling inference tractable. Applied to GBSIEVER-reported LISA Data Challenge sources ($N=2151$ after quality cuts), Galena recovers the disk scale length $R_d=2175^{+51}_{-52}$~pc and scale height $z_d=282^{+7}_{-7}$~pc, consistent with a canonical thin disk, together with the bulge fraction $A=0.187^{+0.011}_{-0.012}$ and bulge scale radius $R_b=773^{+26}_{-25}$~pc; the bulge fraction, now recovered much closer to the literature value of $0.25$ than in our earlier exact-position fit, is measured at $\sim6\%$ statistical precision. An independent particle-swarm optimisation of the same likelihood reproduces these values to within a fraction of a percent, supporting the attribution of the improvement to the Fisher-matrix error propagation rather than to the sampler. The search-stage astrophysical prior improves per-source distance estimates but leaves the hierarchical inference essentially unchanged.
Read the full abstract

All authors Shao-Dong Zhao, Xue-Hao Zhang, Qun-Ying Xie, Yu-Xiao Liu, Soumya D. Mohanty

High Energy Physics (Theory)

The Geometry of Gravitational Radiation

We consider 4-dimensional asymptotically flat vacuum spacetimes near future null infinity endowed with the most general allowable Carroll geometry. We show that the near-boundary radial expansion (to a certain order) can be organised in terms of connections that can be obtained by gauging the conformal Carroll algebra. The only non-vanishing curvatures in this gauging procedure are those that are associated with the special conformal generators and we will refer to these as the $K$-curvatures. The vanishing of these $K$-curvatures defines an asymptotic vacuum spacetime and we use this to construct a boundary (i.e. Carroll) covariant expression for the vacuum (soft) shear in terms of two boundary Carroll scalar fields. The $K$-curvatures transform in a hierarchical fashion into one another under Carroll boosts. This leads to a classification of 4 types of spacetimes: vacuum, strongly and weakly non-radiative, and radiative spacetimes. It is shown that the $K$-curvatures correspond to 5 of the 10 Weyl tensor components at leading order in their $1/r$ expansion. We furthermore observe that one of the $K$-curvatures is equal to the recently found Carroll boost anomaly. Finally, we show that the Bondi loss equations for the boundary energy-momentum-news complex can be cast into a form involving another energy-momentum tensor with vanishing energy flux that is traceless and whose non-conservation is entirely captured by the $K$-curvatures. The BMS currents can be obtained by contracting this latter energy-momentum tensor with a Carroll conformal Killing vector.
Read the full abstract

Computation and Language

Improving Cross-Lingual Token Representations by Adding a Pinch of SALT

Cross-lingual sentence encoders enable scalable transfer across hundreds of languages, powering applications such as translation mining and zero-shot learning in low-resource settings. Although trained for sentence-level alignment, they are increasingly also applied to token-level tasks such as hallucination detection and sequence tagging, exposing a mismatch between training and usage. We propose SALT, a lightweight post-training method that improves token representations by injecting span-level supervision into existing sentence encoders. Across five multilingual token-level benchmarks, SALT achieves the best overall results on four of them, outperforming alternative fine-tuning strategies and competitive encoders. It also improves sentence-level performance on cross-lingual retrieval and classification tasks. These results demonstrate that span-level supervision is an effective signal for improving both token and sentence representations.
Read the full abstract

Complex Variables

Critical cyclicity in Dirichlet-type spaces on the bidisk

Consider the Dirichlet-type spaces on the bidisc defined by $$\mathcal{D}_{\beta}(\mathbb{D}^2)=\Bigg\{f(z_1,z_2)=\sum_{k,l}a_{kl}z_1^kz_2^l\in\mathcal{O}(\mathbb{D}^2): \sum_{k,l\ge 0}|a_{kl}|^2(k+l+1)^{\beta}<+\infty\Bigg\}.$$ Given $\beta_c\in(0,2],$ we construct a function $f$ that belongs to the Dirichlet-type space $\mathcal D_{2}(\mathbb{D}^2)$ of the bidisk and is cyclic in $\mathcal D_\beta(\mathbb{D}^2)$ if and only if $\beta\leq \beta_{c}.$ We also show that the critical index satisfies $\beta_c=2-\mathrm{dim}_H(\mathcal{Z}(f)\cap \mathbb{T}^2),$ where $\mathrm{dim_H}(\mathcal{Z}(f)\cap \mathbb{T}^2)$ is the Hausdorff dimension of the zero set of the function $f$ on the two-torus $\mathbb{T}^2.$
Read the full abstract

Applied Physics

An angular distortion matrix approach for joint wave-speed tomography and aberration correction in scattering media

Despite being one of the most fundamental properties governing wave propagation, the wave-speed distribution is rarely known accurately in reflection imaging. Its estimation relies on analyzing wave distortions undergone by the incident and reflected waves; however, their contribution is difficult to disentangle from the medium reflectivity, particularly in complex media dominated by speckle. Mismatches between assumed and actual wave speeds result in aberrations that degrade the quality of reflectivity images. Yet these aberrations carry information about the underlying wave-speed heterogeneities. Here, we show that an angular distortion matrix, built upon matrix imaging, can unscramble this information to map the wave speed and correct aberrations across the entire field of view. At each point, this matrix isolates the distortions accumulated by the incident and reflected waves along well-defined propagation directions. It reveals strong angular correlations that can be exploited through time-reversal analysis to estimate local phase aberrations, providing observables for wave-speed tomography. The resulting map improves the propagation model used for matrix imaging, after which the process is iterated to refine the wave speed until residual aberrations become negligible. Using ultrasound as proof of concept, we validate the approach in a tissue-mimicking phantom and illustrate its clinical potential in vivo for breast and liver imaging. Sound-speed maps show contrast consistent with a malignant breast lesion and expected values in a healthy liver, while aberration correction sharpens reflectivity images routinely interpreted by clinicians. Beyond ultrasound, the approach extends naturally to any wave modality in which reflection-matrix imaging can be implemented.
Read the full abstract

All authors Chenyu Cui, Flavien Bureau, Foucauld Chamming's, Alexandre Aubry, Naiara Korta Martiartu

Cosmology and Nongalactic Astrophysics

Learning cosmic web environments with diffusion models

The cosmic web, consisting of an intricate network of voids, walls, filaments, and nodes, encodes key information about structure formation and the cosmological parameters that govern it. In the era of high-precision cosmology, large ensembles of numerical simulations are required to analyse next-generation galaxy surveys, motivating the use of generative models to circumvent their high computational cost. While they have shown promise in emulating high-fidelity cosmic web simulations, their ability to capture distinct cosmic web environments remains largely unexplored. For this study, we trained a diffusion model on the Quijote N-body simulation suite to investigate the semantic information learnt by its self-attention maps. Using statistical estimators such as the Dice coefficient and cross-power spectra, we quantified the correspondence between attention maps and cosmic web environments defined by the T-Web classifier. We find that attention maps of varying spatial resolutions across different layers capture overdense and underdense structures in distinct ways, exhibiting strong positive correlations and anti-correlations with both the overall matter distribution and individual cosmic web environments. Moreover, the diffusion model predominantly encodes cosmological information at intermediate-to-large spatial scales, indicating that attention maps primarily capture globally coherent structures. Our results show that, beyond accurately reproducing two-point statistics, diffusion models learn a multi-scale representation of the cosmic web through self-attention, including non-Gaussian information.
Read the full abstract

All authors Mehdi Noor, Tony Bonnaire, Nabila Aghanim, Aurélien Decelle

Computer Vision

Vague2Detect: Handling Ambiguous Prompts in Knowledge-Based Open-World Detection

Real-world detectors must often interpret functional or ambiguous prompts, yet conventional models such as YOLO remain restricted to fixed class lists. Even open-vocabulary models like YOLO-World frequently misalign vague language with the intended objects. Building on our prior work Commonsense-Guided Open-World Object Detection Using LLMs and Visual-Semantic Matching, we address YOLO-World's limitations in grounding task-driven queries. We propose Vague2Detect, a hybrid pipeline in which a fine-tuned Sentence-BERT retrieves candidates from a structured household Knowledge Base (KB), and YOLO-World verifies their presence in the image. For prompts outside the KB, a large language model (GPT-3.5-turbo) generates candidate descriptions, dynamically expanding the KB to cover novel concepts. On a benchmark of household scenes using custom images and an Open Images V7 subset, YOLO-World alone achieves only 32% Vague Prompt Success Rate (VPSR), the ability to map ambiguous queries to correct detections. In contrast, Vague2Detect improves performance to 61% VPSR with high precision, and up to 85% when augmented with GPT fallback.
Read the full abstract

Sound

Robust Rank Aggregation for Multimodal Speech-Based Alzheimer's Disease Detection

Speech-based Alzheimer's disease (AD) detection has recently benefited from multimodal foundation-model representations that integrate complementary acoustic and linguistic information. However, conventional probability averaging over these complementary classifiers is unreliable, because their posterior probabilities exhibit mismatched scales: identical values may reflect different confidence levels across models. We propose a robust rank aggregation framework that aggregates normalized prediction ranks instead of posterior probabilities. Each subject is scored by its percentile within a fixed training-cohort distribution of out-of-fold predictions; since rank ordering is invariant to monotonic transformations, this avoids probability-scale mismatch while preserving classifier confidence ordering. A confidence-gated Random Forest further corrects residual errors using clinically interpretable linguistic features, overriding the rank prediction only when the two disagree and the RF is highly confident, without additional deep model training or explicit posterior-probability calibration. On ADReSS2020 and ADReSSo2021, the method achieves accuracies of 95.83% and 90.14%, respectively, comparing favorably with previously reported results.
Read the full abstract

Audio and Speech Processing

SpeechAnnotator: A Context-Aware Multi-Agent Framework and Benchmark for Multidimensional Speech Annotation

Recent controllable speech generation requires training data with fine-grained annotations of speaker traits, prosody, emotion, paralinguistic cues, acoustic scenes, and context. Existing workflows often rely on manual correction, paid hosted multimodal services, or fixed processing chains, which limits large-scale data processing through annotation cost, external-service dependence, or weak cross-stage recovery. We introduce SpeechAnnotator, a locally deployable, context-aware multi-agent framework built entirely from open-source models and tools. Supporting frontend modules first obtain speaker-aware segments and final segment transcripts, while prior evidence extractors attach heterogeneous segment-level cues. Three specialist agents then collaborate through shared state: the Planning Agent converts local audio evidence, speaker history, neighboring segments, and recording-level context into field-specific contracts; the Labeling Agent performs contract-guided multimodal prediction for directly observable attributes; and the Review Agent runs a bounded review loop that checks evidence support and cross-segment consistency, triggering relabeling only for unsupported or inconsistent fields. To address the fragmentation of existing evaluation resources across isolated tasks and narrow-domain test sets, we introduce SpeechAnnotator-Bench (SA-Bench), containing 8.87 hours of human-annotated audio across nine source formats, together with SpeechAnnotator-Eval (SA-Eval), which separates Timeline-Eval for speaker-aware timeline recovery, Closed-Eval for finite-set attributes, and Open-Eval for open-ended attributes. Experiments and ablations show that SpeechAnnotator provides a locally deployable alternative to commercial audio-capable systems, while the bounded review loop improves multidimensional annotation through evidence- and context-aware field-level recovery.
Read the full abstract

All authors Qirui Zhan, Shuiyuan Wang, Jingbin Hu, Haoyu Zhang, Xiaming Ren, Jinrui Liang, Chaoren Yu, Bengu Wu, Yunxiang Chen, Houdun Liu, Su Feng, Liumeng Xue, Lei Xie

Optimization and Control

Boundary control for optimal mixing by two-dimensional second-grade fluids

We study optimal boundary mixing of a nondiffusive scalar transported by a two-dimensional incompressible second-grade fluid. The control is the tangential traction in a Navier-slip boundary condition, and the principal objective is the terminal $(H^1(\Omega))'$ mix-norm, supplemented by a quadratic control cost and an optional enstrophy reward. The second-grade constitutive law introduces a spatially filtered acceleration and a generalized vorticity; after lifting the nonhomogeneous boundary data, both $\mathbf{g}$ and $\partial_t\mathbf{g}$ enter the state equation, where $\mathbf{g}$ is the boundary control variable. We prove global state well-posedness for a passive scalar and well-posedness on a common, control-independent local interval for an active scalar. We then establish existence of an optimal control, directional differentiability of the control-to-state map in the topology required by the terminal objective, and a weak backward adjoint formulation. A duality identity yields the first-order variational inequality. In the absence of the friction coefficient $\beta$ and enstrophy reward weight $\zeta$ (that is, $\beta=\zeta=0$), the additional adjoint regularity implies uniqueness of the optimal control for all sufficiently large control penalties $\gamma$.
Read the full abstract

Machine Learning

Adversarial Training for Tabular Credit Scoring: A Multi-Attack Robustness Evaluation in P2P Lending

Machine learning-based credit scoring is increasingly central to Peer-to-Peer (P2P) lending, yet its resilience to adversarial manipulation, where applicants strategically alter self-reported inputs to secure favourable decisions, remains poorly understood. Most adversarial-robustness evidence comes from image and text domains and evaluates a single attack against a matching defence, offering little guidance on how defences generalise across attack types in tabular credit data. We address this with a systematic train-test robustness benchmark on a large Lending Club subset, spanning three model families (logistic regression, a feed-forward neural network, and a transformer for tabular data) and four attacks confined to applicant-mutable features: Fast Gradient Sign Method (FGSM), Projected Gradient Descent (PGD), Salt-and-Pepper (S&P) noise, and DeepFool, plus a mixed-attack regime. Across a full grid evaluated with stratified cross-validation, adversarial training sharply improves robustness against the attack it is trained on and transfers well within the gradient-based family, but transfers weakly to non-gradient corruption, so single-attack defences overstate real-world resilience. Mixed training delivers the most balanced robustness across heterogeneous attacks while preserving clean-test performance, supporting multi-attack stress testing in credit-model governance.
Read the full abstract

All authors Gijs A. F. Niewzwaag, Marijn G. S. Veth, Manuele Massei, Marcos R. Machado