Peer Revenue Returns subject to peer review

Papers

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

All of arXiv

Tuesday 1 September

Image and Video Processing

Prior-Guided Implicit Neural Representations for Single-Subject Diffusion MRI Super-Resolution

Resolving complex fiber geometries in brain white matter requires high-resolution diffusion MRI at the cost of long acquisition times. This leads many clinical protocols to opt for low-resolution scans, making downstream microstructure estimation and tractography challenging. Implicit neural representations (INRs) can model the diffusion signal continuously, enabling native single-subject super-resolution by querying the network at arbitrary spatial coordinates, yet existing methods often suffer from long training times and lack a mechanism to incorporate anatomical priors to regularize super-resolution by constraining the space of plausible reconstructions. To address these limitations, we propose a novel transfer-learning framework that pre-trains an INR on a high-resolution template and then adapts it to subject-specific scans via registration and fine-tuning. For $4\times$ through-plane super-resolution from 5 mm to 1.25 mm on Human Connectome Project (HCP) data, our method reduces NRMSE by 36-49% and increases FSIM by 24-43% over a recent baseline with $6\times$ faster training, outperforming competing INR-based methods across both image quality and domain-specific metrics. Code is available on the project page at https://abdulkaderghandoura.github.io/research/msc-thesis/ .
Read the full abstract

All authors Abdulkader Ghandoura, Marsil Zakour, William Consagra, Yogesh Rathi

General Relativity and Quantum Cosmology

Near-extremal asymptotics and strong cosmic censorship for black holes immersed in a Chaplygin-like dark fluid

We investigate strong cosmic censorship (SCC) for massless scalar perturbations of electrically neutral black holes immersed in a Chaplygin-like dark fluid (CDF). The matter distribution produces an asymptotically de Sitter exterior while supporting an inner Cauchy horizon. We derive a closed-form parametrization of the extremal and Nariai horizon boundaries. This analytic control yields explicit near-extremal asymptotics for the horizon splitting and Cauchy-horizon surface gravity, together with an analytic expression for the leading near-extremal scalar quasinormal spectrum. The fundamental near-extremal damping rate approaches the Cauchy-horizon surface gravity, whereas higher angular multipoles retain explicit dependence on the CDF extremal geometry. Combining these analytic results with global quasinormal-mode calculations, we determine the spectral gap and assess the Christodoulou formulation of SCC throughout the three-horizon domain. Potential SCC violation is confined to a narrow region near extremality. As the effective cosmological scale increases, the lower boundary of this region is controlled successively by de Sitter, photon-sphere, and de Sitter modes, while its normalized width varies nonmonotonically.
Read the full abstract

All authors Hao-Peng Yan, Zeng-Yi Zhang, Xiang-Qian Li, Xiao-Jun Yue

Optics

Topological charge switching in trapped polariton condensates

Topological charges in photonic systems provide a robust degree of freedom with direct applications in optical information processing. Bound states in the continuum (BICs) in photonic crystal slabs inherently carry such quantized topological charges, yet they are topologically protected, making active reconfiguration fundamentally challenging. Here, we demonstrate topological charge switching in trapped BIC polariton condensates in a photonic crystal, where spatial modulation of the photonic crystals supports multiple confined states, each carrying a distinct charge. Trapped condensates within the Dirac bandgap are expelled into the leaky band via polariton blueshift, enabling single-mode selection by solely tuning the excitation power. This work offers a platform for optically controlling topological charges by harnessing polariton nonlinearity.
Read the full abstract

All authors Jaewon Kim, Hyun Gyu Song, Daegwang Choi, Yong-Hoon Cho

Algebraic Geometry

Towards Categorical Kähler Geometry

We outline the contours of an emerging theory of Kähler metrics in derived noncommutative geometry. This is a refinement of the theory of Bridgeland stability conditions encoding underlying differential-geometric structures. We propose elements of such a structure in both Archimedean and non-Archimedean settings, including metrized objects, mass measures satisfying a BPS inequality, harmonic metrics, minimizing flows, and complexified Kähler potentials. We develop the framework through examples and constructions involving Fukaya categories, quiver representations and associated C$^*$-algebras, spectral networks, and comonadic adjunctions of stable $\infty$-categories.
Read the full abstract

All authors Fabian Haiden, Ludmil Katzarkov, Maxim Kontsevich, Pranav Pandit

Quantum Physics

Information geometry of non-equilibrium quantum states: Mixed metric structures on an extended information manifold

We discuss an information-geometric framework for characterizing quantum states in non-equilibrium dynamics. Using the transverse-field Ising chain model as a laboratory, we investigate the quantum Fisher information metric with particular emphasis on the mixed metric component $g_{ht}$ and related geometric observables, where $h$ is a controllable post-quench Hamiltonian parameter and $t$ is real time. The framework treats $h$ and $t$ as coordinates of an extended information manifold $(h,t)$. The geometric observables characterize the speed of evolution on the manifold and the alignment between the temporal and parameter-deformation directions. Correlations between these geometric quantities and two widely used measures of non-equilibrium dynamics, the entanglement entropy and the Loschmidt echo, are analyzed.
Read the full abstract

Geometric Topology

Lipshitz-Sarkar refines mirrors more than Khovanov

Lipshitz and Sarkar asked in an ICM 2018 question whether their homotopy-theoretic refinement of Khovanov homology can detect mirror reflection when ordinary integral Khovanov homology cannot. We give an affirmative answer by presenting a prime knot whose integral Khovanov homology agrees with that of its mirror in every bidegree, including torsion, and yet over $\mathbb{F}_2$, the second Steenrod square has rank one for the knot and rank zero for its mirror in a specified bidegree. We also give a composite example. The construction, search and computations are done with the help of chat-GTP 5.6 sol Pro combing over the Regina Census of knots.
Read the full abstract

Mesoscale and Nanoscale Physics

Topological charges and parity selection at Floquet quasienergy degeneracies

The quasienergy spectrum of a strongly driven two-level system as a function of the driving parameters exhibits conical intersections, which are enabled by hidden time-nonlocal symmetries. We show that each such crossing carries a quantized topological charge: the Floquet–Berry phase acquired along an adiabatic loop around a cone is equal to a $\mathbb{Z}_2$-valued charge. We further identify a second family of degeneracies that occurs at vanishing driving amplitude, when the level splitting matches $m$ energy quanta of the field. Along the Stark-shifted resonance line, the minimum quasienergy gap opens as $|A|^m$, and the charge is nontrivial only for odd $m$. We analytically derive both results from a perturbative reduction to a spin-$1/2$ in an effective two-dimensional magnetic field and confirm them numerically through the Bargmann invariant. Moreover, we propose a chirality-based protocol that cancels the dynamical phase to isolate the geometric one, and an ancilla-based Ramsey readout that renders the topological charge directly observable.
Read the full abstract

Algebraic Geometry

On Algebraic Integrability of Vector Fields with Rational Function Coefficients that Separate Variables

We work over a field of characteristic zero - primarily over an algebraic closure of the field of rational numbers. We prove a necessary and sufficient condition for a vector field whose coefficients are rational functions separating variables to be algebraically integrable, that is, the subring of rational functions killed by this vector field, its ring of first integrals, is of maximal possible dimension. We do it arithmetically by reducing the vector field modulo almost all primes. In particular, we verify the generalized Grothendieck–Katz p-curvature conjecture for foliations defined by these vector fields. Finally, we use the outcome of that verification to provide explicit formulas for coefficients of all algebraically integrable vector fields separating variables, and their first integrals.
Read the full abstract

Analysis of PDEs

The Vlasov-Maxwell point charge system : study of the spherically symmetric case

We consider the Vlasov-Maxwell system with a repulsive point charge. Under a spherical symmetry condition, this reduces to the relativistic Vlasov-Poisson system with a point charge. For this system, the linear characteristics solve the equations of motion for the relativistic Kepler problem. We define action-angle variables for the linearized equation, and then study the nonlinear system in these coordinates. Within this framework, we prove global existence of classical solutions for small initial data and establish their modified scattering behavior. The proof makes use of the ideas developed in the non-relativistic case.
Read the full abstract

Instrumentation and Detectors

Time performance of USTC-IME LGAD under synchrotron light source focused X-ray

The time performance of Low Gain Avalanche Diodes (LGADs), designed by the University of Science and Technology of China (USTC) and fabricated by the Institute of Microelectronics of Chinese Academy of Sciences (IME), was characterized at the Shanghai Synchrotron Radiation Facility (SSRF). The experiment was conducted at the BL16B1 beamline, which delivers a focused X-ray beam with a diameter of 500 \textmu\text{m}, a repetition period of 2 ns, and a photon energy of 10 keV. Using a fast oscilloscope, waveforms containing continuous signal pulses were recorded within a 50 ns time window. The LGAD under test successfully resolved the 2 ns period of the SSRF. To mitigate pile-up effects and extract pulse-by-pulse information from the acquired waveforms, a waveform-level global template fitting method was employed. The time resolution was then estimated using a combined profile likelihood approach, with a restricted maximum-likelihood (REML) correction accounting for the per-waveform degrees of freedom, yielding a value of 136.8 $\pm$ 0.15 (stat.) $\pm$ 11.9 (syst.) ps. The complete reconstruction and fitting chain was validated end-to-end using simulated waveforms in which the timing deviations of individual large pulses were drawn from Gaussian distributions with fixed standard deviations. The effect of random photon absorption depth on the time resolution of LGADs was studied through dedicated simulations.
Read the full abstract

All authors Tianao Wang, De Zhang, Zheng Liang, Chaozhi Li, Jiaqi Luo, Yu Nie, Tianshuo Yue, Zebo Tang, Hao Liang, Tongpu Yu, Wanbing He, Yanwen Liu

Statistical Mechanics

Kibble–Zurek Mechanism and Defect Freezing in Imbalanced-Pairing Kitaev Models

We investigate driven dynamics across critical and exceptional points in the one- and two-dimensional imbalanced-pairing Kitaev models using both the wave-function normalization approach and the biorthogonal framework. For a positive pairing imbalance parameter, the quasiparticle spectrum remains real, and a pairing imbalance neither shifts the equilibrium phase boundaries nor generates imaginary eigenenergies. In this regime, the defect density follows the conventional Kibble–Zurek scaling in one dimension and the extended Kibble–Zurek scaling, arising from a gapless manifold, in two dimensions within both frameworks. The corresponding scaling exponents are therefore governed by those of the Hermitian transition. For a negative pairing imbalance parameter, time-reversal symmetry is broken, the quasiparticle spectrum develops complex eigenvalues, and the gap closes at exceptional points. For ramps ending at an exceptional point, the defect density follows the modified Kibble–Zurek scaling in the wave-function normalization approach, whereas it obeys the conventional Kibble–Zurek scaling in the biorthogonal framework. When the ramp traverses the time-reversal-symmetry-broken region, a finite density of defects remains even in the adiabatic limit, leading to defect freezing in both frameworks. Although this frozen background indicates a breakdown of adiabaticity, the excess defects generated on top of this background continue to obey the conventional Kibble–Zurek scaling in one dimension and the extended Kibble–Zurek scaling in two dimensions.
Read the full abstract

All authors R. Jafari, Alireza Akbari, Shukhrat Mardonov, A. Langari

Functional Analysis

The unique predual problem for Lipschitz spaces, revisited

In his 2018 paper “On the unique predual problem for Lipschitz spaces”, N. Weaver published proofs that Banach spaces $\mathrm{Lip}_0(M)$ of Lipschitz functions on a complete metric space $M$ have strongly unique preduals whenever $M$ has finite diameter or is geodesic. A gap was recently noticed in the proof of a crucial lemma that claimed that the property of having a strongly unique predual passes to 1-codimensional weak$^*$-closed subspaces. In this note, we confirm that the lemma is actually false by providing an explicit counterexample. We also expand on some of Weaver's original arguments to provide a new, valid proof of the following particular case: $\mathrm{Lip}_0(M)$ has a strongly unique predual whenever $M$ is a convex subset of a finite-dimensional normed space.
Read the full abstract

Digital Libraries

The zbMATH Open Knowledge Graph: Tracing Centuries of Mathematical Research

We present the zbMATH Open Knowledge Graph, a large-scale RDF knowledge graph (KG) covering more than 250 years of mathematical scholarship. Unlike existing scholarly knowledge graphs that primarily capture bibliographic metadata and citation structures, the zbMATH Open KG integrates expert-curated semantic content, including reviews, keywords, subject classifications, software references, and disambiguated authorship. This combination of domain-specific representation of mathematical knowledge and extensive temporal coverage supports analyses that require fine-grained exploration of mathematical concepts, research fields, and scholarly relationships over time. The resulting graph comprises 34 million entities and 168 million RDF triples represented using established Semantic Web vocabularies, supporting interoperability and FAIR data principles. We further demonstrate its capabilities through query-driven historically grounded scholarly exploration use cases, illustrating how the knowledge graph can surface relationships and patterns that may be difficult to identify from bibliographic and citation information alone. The zbMATH Open KG provides an open semantic infrastructure for studying the development of mathematical knowledge and tracing scholarly connections across centuries of scholarship.
Read the full abstract

Computer Vision

ReFlowSET: Representation-Aligned Latent Flow Matching for SAR-to-EO Image Translation

SAR-to-EO image translation aims to generate electro-optical (EO) imagery from synthetic aperture radar (SAR) observations. Existing latent diffusion approaches typically inherit a predetermined autoencoder, although reconstruction fidelity can vary substantially across codecs and modalities. Because the latent codec affects the round-trip preservation of both SAR conditions and EO targets, codec selection constitutes a fundamental design choice; nevertheless, existing methods largely rely on codecs pretrained on natural images. To remedy this, we introduce ReFlowSET, a conditional latent flow-matching framework that selects its codec through a joint SAR–EO reconstruction audit. Rather than inheriting a heavyweight pretrained generator, ReFlowSET trains a substantially smaller conditional DiT from scratch in the selected latent space, using dual-stream SAR conditioning followed by joint feature refinement. To provide semantic guidance for this from-scratch training, intermediate noisy-EO features are aligned with clean target-EO representations extracted by a frozen vision foundation model. This alignment is used only during training and introduces no additional inference cost. Experiments on QXS-SAROPT and SAR2Opt demonstrate state-of-the-art performance across diverse perceptual fidelity and distributional metrics. Code and pretrained weights are publicly available at https://github.com/KAIST-VICLab/ReFlowSET.
Read the full abstract

Artificial Intelligence

CoBRA: Learning Tool-Use Boundaries via Counterfactual Margins

As large language models increasingly act through external tools, deciding when to call a tool has become a central problem alongside deciding how to use it. Unnecessary tool calls introduce latency, cost, retrieval noise, and error propagation, while missed calls hurt knowledge-intensive queries or questions requiring up-to-date evidence. Existing methods typically trigger tools from absolute query or generation signals, such as difficulty, confidence, or final task reward, and therefore lack an explicit estimate of the instance-level marginal benefit of tool use. We propose CoBRA, a counterfactual boundary-learning framework for tool-augmented language models. CoBRA first constructs internal and external experts from the same base model, collects paired trajectories, and estimates the reward margin between answering with and without tools. This margin partitions data into internal-favored, external-favored, and ambiguous cases. CoBRA then uses clear-margin samples for Boundary-Aware Cold-Start SFT, followed by MARS-RL with reference-split rollouts and counterfactual marginal advantages to optimize boundary decisions. Experiments with retrieval as the main tool on Qwen3-4B show that CoBRA improves tool-use efficiency and boundary-sensitive answer accuracy while maintaining strong performance on tool-dependent out-of-distribution questions.
Read the full abstract

All authors Wenhao Zou, Xianglong Liu, Wendong Bi, Hanjie Wang, Simin Zhao, Gong Zhi

Sound

ABSE-NET: A Lightweight Neural Model for Active Binaural Speech Enhancement in Open-Fit Hearing Aids

Open-fit hearing aids have attracted growing attention due to their superior wearing comfort. However, the open-fit design inevitably causes acoustic leakage into the ear canal, degrading the performance of existing binaural speech enhancement (BSE). To this end, we propose ABSE-NET, an active BSE framework integrating active noise control (ANC) with BSE to jointly enhance target speech and suppress acoustic leakage. The ABSE-NET pipeline cascades a binaural MVDR (BMVDR) with a lightweight neural network (LNN). The former achieves a coarse BSE, whereas the latter simultaneously cancels acoustic leakage and compensates for BMVDR-induced distortion. The LNN uses an encoder-decoder with a feature fusion module, which includes frequency-time dependency learning and convolutional attention blocks. Unlike traditional BSE+ANC solutions via adaptive filtering, ABSE-NET needs no in-ear microphone in practical deployment. Experiments validate its superiority over state-of-the-art methods. Code repository: https://github.com/Bream101/ABSE-NET.
Read the full abstract

All authors De Hu, Xue Du, Qingying Zhao, Qintuya Si

High Energy Astrophysical Phenomena

The Radius of the Neutron Star PSR J0614-3329 from NICER Data

Neutron star radius measurements, particularly using X-ray data collected with the Neutron star Interior Composition Explorer (NICER), have provided invaluable information for models of cold, catalyzed matter at densities above that of nuclear saturation. Here we present an analysis of NICER data on the 318-Hz pulsar PSR J0614-3329, which has a mass ~1.4-1.5 solar masses determined from radio observations. The best-fitting model we explore has three uniform-temperature circular hot spots and yields a symmetric 68% credible range for the equatorial circumferential radius of 9.88-12.77 km. We also explore joint fits to the NICER data and the X-ray Multi-Mirror (XMM-Newton) data on this pulsar, but find that the models that best fit both data sets systematically underpredict the XMM-Newton data when, as is standard, it is assumed that the XMM-Newton background is known from observations of surrounding fields. Finally, we discuss the implications of our results, combined with data on other neutron stars, for the properties of the dense matter in neutron star cores.
Read the full abstract

All authors M. C. Miller, A. J. Dittmann, I. M. Holt, F. K. Lamb, C. Chirenti, Z. Arzoumanian, J. Berteaud, S. Bogdanov, K. C. Gendreau, A. K. Harding, W. C. G. Ho, C. Kalapotharakos, S. M. Morsink, P. S. Ray, R. A. Remillard, Z. Wadiasingh, M. T. Wolff

Superconductivity

Resolving unconventional gap structure in kagome superconductors with hybrid microwave circuits

Unconventional superconductivity is a hallmark of exotic quantum matter, where determining the pairing symmetry is essential for uncovering its microscopic origin. Kagome superconductors provide a fertile landscape for emergent phenomena arising from strong electronic correlations and nontrivial band topology, yet their superconducting pairing symmetry remains elusive. The superconducting gap structure is commonly probed via electrodynamic response, but such measurements are inapplicable to thin flakes due to their small volume and delicate nature. Superconducting microwave resonators offer a coherent and highly sensitive platform for probing electrodynamic responses, with versatile designs that enable incorporation of diverse materials and geometries. Here, we integrate flakes into microwave circuits, enabling noninvasive access to the superfluid response through contactless coupling that preserves structural integrity. By engineering the device geometry to suppress parasitic two-level-system losses that dominate dissipation in microwave circuits, we isolate the intrinsic material response. Remarkably, the temperature-dependent superfluid density exhibits linear behavior at low temperatures, consistent with a nodal gap structure. Our approach establishes a noninvasive platform for probing electrodynamic response of fragile superconducting flakes while advancing hybrid microwave architectures for quantum technologies.
Read the full abstract

All authors Yejin Lee, Haolin Jin, Sushmita Chandra, Berit H. Goodge, Edouard Lesne, Tommaso Confalone, Francesco Tafuri, Davide Massarotti, Golam Haider, Kornelius Nielsch, Bernd Büchner, Claudia Felser, Debanjan Chowdhury, Nicola Poccia, Uri Vool

High Energy Astrophysical Phenomena

Connecting radio pulsars, magnetars, and XDINSs in a unified evolutionary framework using simulation-based inference

Understanding the Galactic population of isolated neutron stars within a unified framework provides key insights into their birth properties, evolutionary pathways, and the connections between different neutron star classes. In this work, we aim to reproduce both the observed radio and quiescent X-ray emission from isolated neutron stars, including radio pulsars, magnetars and X-ray dim isolated neutron stars (XDINSs). We develop a comprehensive population synthesis framework that models neutron star birth properties, their dynamical, rotational, and magneto-thermal evolution, as well as radio and X-ray emission, and the selection effects of corresponding surveys. To simulate realistic X-ray spectra, we account for magnetospheric resonant cyclotron scattering and interstellar absorption. Additionally, we model the observational bias introduced by magnetar outbursts, by linking the outburst rate to magnetic stresses in the stellar crust. We then employ a simulation-based inference method, namely truncated sequential neural posterior estimation, to reconstruct the birth properties, such as the initial magnetic field distribution. We find that the Galactic neutron star population can be described by a two-component log-normal distribution of birth magnetic fields with peaks at $5 \times 10^{12}$ G and $10^{14}$ G, respectively. We further find that the Galactic neutron star birth rate is around 3-4 per century. Our results help establish the contribution of neutron stars as central engines of powerful astrophysical transient phenomena, such as fast radio bursts, super-luminous supernovae and gamma ray bursts.
Read the full abstract

All authors Michele Ronchi, Celsa Pardo-Araujo, Vanessa Graber, Nanda Rea, Clara Dehman, Davide De Grandis, Alessio Marino, Francesco Coti Zelati, Joeri van Leeuwen, José A. Pons

Artificial Intelligence

Few-Shot Out of Domain Intent Detection with Covariance Corrected Mahalanobis Distance

Conversational agents like chatbots and voice assistants are trained to understand and respond to user intents. On encountering an utterance with an intent different from the ones they have been trained on, these agents are expected to classify the intent as `unknown' or `out of domain'. This problem is known as out of domain (OOD) intent detection. Podolskiy et al. (2021), showed that Mahalanobis distance can be used effectively for identifying OOD intents, outperforming competing approaches. However, their method fails to outperform the baselines in the practically important few-shot setting. In this paper we analyze the reason for low performance and propose a covariance corrected Mahalanobis distance for detecting out-of-domain intents.
Read the full abstract

Computer Vision

Conditional Flow Matching for Cross-Field MRI Harmonisation

Magnetic resonance images of the same subject look markedly different across field strengths, which complicates the comparison and pooling of data across sites. We address cross-field brain-MRI translation for the MRIxFields2026 challenge, and in particular its Task~3: a single model that translates between any directed pair of the five field strengths and across three contrasts. We phrase the problem as a conditional flow matching path: because the source and target volumes are spatially registered, we learn a velocity field that carries the source slice directly to the target slice, rather than starting from noise. To learn this mapping from only three paired subjects, the unified model is trained in three stages: a degradation-bridge pretraining that distills a restoration prior from the abundant unpaired retrospective cohort, a cross-field finetuning over all directed pairs on the paired cohort, and an adversarial refinement that sharpens the output. At inference, we integrate the learned velocity with a second-order Heun solver in a handful of steps. A restoration prior learned without any paired data already reaches a mean SSIM of 0.837, and each subsequent training stage improves on it. A single 6.3M-parameter model thereby covers all 60 field-pair and contrast combinations, with inference in five solver steps per slice. On the challenge evaluation set the model reaches a mean SSIM of 0.909, averaged over the three contrasts, outperforming regression and diffusion baselines built on the identical network on all three challenge metrics.
Read the full abstract

All authors Baris Imre, Aram Salehi, Levente Baljer, Andrew Webb, Marius Staring, Efe Ilicak

Computer Vision

Candidate-Expanding Routing with Permutation-Stabilized Experts for Mixed-Format Medical VQA

Mixed-format medical visual question answering (VQA) requires stable option selection and machine-readable free-text output. The two formats fail differently: multiple-choice predictions can change with option symbols or positions, while clinically plausible open answers can fail automated evaluation when serialization is malformed. We address both challenges with an answer-text memory, a permutation-stabilized vision–language expert, and a sparse candidate- expanding router. The cyclic schedule follows prior work; our contribution is to make expert top-2 a routable candidate alongside memory and expert top-1. On a 1,403-case retrospective internal analysis, this expansion improves a matched binary router from 88.95% to 91.73% (+2.78 percentage points; 95% CI 1.57–3.99), with 56 rescued errors and 17 regressions. Oracle coverage rises from 90.31% to 96.15%, and the final submitted configuration reaches 92.23% on the same retrospective split. For open questions, strict generation and deterministic guards produce 475/475 schema- valid participant-facing outputs without repair, retry, or hard-gate failure. Visual ablations reveal substantial textual dependence. Candidate expansion supplies the principal controlled routing gain; open-path evidence establishes output-contract validity rather than clinical correctness in medical use or deployment.
Read the full abstract

Computation and Language

PersianAnonymizer: Evaluating LLM-Labeled Training for Efficient NER-based Anonymization in Persian

We target practical anonymization of Persian customer chats by training a compact NER model from LLM-labeled supervision and selecting the best labeler for deployment. We compare three instruction-tuned LLMs: DeepSeek-V3-0324, GPT-OSS-120B, and Qwen3-235B-A22B-Instruct-2507, to produce span annotations under a shared JSON protocol, yielding four corpora (OSS_ZeroShot, Qwen_ZeroShot, Qwen_FewShot, DeepSeek_FewShot). A MatinaRoberta-based token-classifier is trained per corpus and evaluated with token-level Precision/Recall/F1 (overall and per-class). We also report Label Coverage Recall (LCR), the proportion of gold non-O tokens predicted as non-O, and quantify cross-labeler behavior via a token-level Venn on test annotations. Finally, we contrast test-set annotation latency of the LLMs on H200 nodes with the trained NER's test-time labeling on a single RTX 3090. Results show that supervision from OSS_ZeroShot yields the strongest macro-F1 and LCR, while the resulting NER labels an entire 40K-message test set in approximately 2 minutes on one consumer GPU. This establishes a practical path to high-quality, low-cost anonymization for Persian industrial data.
Read the full abstract

Image and Video Processing

Learning the Shoreline: A Very High-Resolution Approach to Reef Island Dynamics

Pacific atoll islets are often described as stable in global-scale studies, typically based on long-term shoreline proxies such as vegetation line or morphometrics like planform surface area. While informative, these approaches can obscure short-term, localized coastal dynamics -including changes in island shape and position -that are critical for ecosystem function, cultural practices, and coastal infrastructure resilience. This study presents a transferable, automated approach to shoreline monitoring using very high-resolution Pléiades imagery and a XGBoost classifier. The method integrates spectral indices and textural features to delineate the outer limit of emerged land, including vegetated areas, beaches, man-made surfaces, and beach rock. This shoreline definition supports finescale, spatially explicit monitoring of reef island dynamics, even in morphologically complex environments. Developed and tested on multiple atolls in French Polynesia (Tetiaroa, Tikehau, Hao, and Puka Puka), the model achieves high accuracy (mean Intersection over Union $\approx$ 0.99; Mean Absolute Positional Error $\approx$ 1.28 m) and demonstrates strong performance on both training and held-out sites, validating its spatial transferability. The extracted shorelines reveal subtle but significant island-scale changes in extent, configuration, and spatial position that remain undetected by conventional shoreline proxies and surface metrics. By enabling highprecision, scalable shoreline monitoring, this method provides a more nuanced understanding of atoll change processes. It supports Pacific efforts to move beyond narratives of passive loss toward frameworks of resilience and adaptation, while providing spatial tools tailored to low-lying island realities.
Read the full abstract

Computer Vision

PredErase: Training-Free Object-and-Effect Removal with Predictive Latent Guidance

Removing an object is not the same as filling its mask. Cast shadows and contact shading usually lie outside the user-provided instance mask M_obj, so a frozen Fill model that edits only that mask leaves the object's photometric footprint on nearby surfaces. Supervised removers learn this joint erasure from paired clean plates. Training-free editors freeze pretrained weights, yet most still treat M_obj as the entire editable support and steer sampling with CLIP or DINO energies that do not predict the occluded scene. We present PredErase, a training-free inference procedure on frozen FLUX.2 and I-JEPA. The method separates where Fill may rewrite pixels from what structure should occupy the hole. A contact-band expansion M_flux of M_obj exposes local residuals on the supporting plane. I-JEPA, pretrained for masked token prediction, supplies a context-conditioned hole target in representation space; sparse projected gradients align decoded Fill completions with that target inside the instance, while coordinates outside the packed support stay locked. Under instance-only masks on RemovalBench, RORD-Val, and DEFACTO-Val, PredErase improves the native FLUX.2 backbone. Supervised removers remain stronger on several full-image appearance metrics; the supported claim is training-free object-and-effect editing of frozen Fill, not replacement of paired-data erasers.
Read the full abstract

All authors Waikit Xiu, Qiang Lu, Junbiao Chen, Xiying Li

Computer Vision

ASSERT: Adaptive Stochastic Sampling for Robust Diffusion Models on Analog Compute-in-Memory Hardware

Diffusion models achieve strong image generation quality but incur high iterative denoising costs. Analog compute-in-memory (CIM) can accelerate matrix-vector multiplications, yet spatial memory variations perturb weights and accumulate during sampling. Unlike conventional neural networks, diffusion models' temporal sensitivity to hardware noise remains underexplored. We investigate diffusion inference using a noise model calibrated and validated against measurements collected from multiple physical CIM chips. Our results show that the early, high-noise denoising stage is substantially more vulnerable than the final refinement stage. A first-order trajectory analysis attributes this behavior to the repeated propagation of correlated prediction errors induced by a fixed hardware mapping. Based on this observation, we propose ASSERT, a training-free sampler that uses higher stochasticity early and smoothly transitions to deterministic denoising. The injected stochasticity changes subsequent activation trajectories and thereby reduces their alignment with persistent spatial errors. Across the evaluated settings, ASSERT achieves up to 2.58$\times$ lower FID than deterministic DDIM on high-resolution datasets and 7.68$\times$ lower FID in the CIFAR-10 step-count study, without changing model parameters or the number of network evaluations.
Read the full abstract

All authors Yuannuo Feng, Yizhe Chen, Wenshuai Yao, Yuxin Xie, Ngai Wong, Wenyong Zhou, Wang Kang

Cryptography and Security

Influence of Logging Frameworks on Bind9

Host-based Intrusion Prevention Systems (IPS) rely on application logs to detect and block malicious activity. However, on modern high-speed networks the logging subsystem itself becomes a bottleneck: an attacker can hide traces simply by generating enough traffic to overwhelm the application's log pipeline, dropping crucial traces. In this work, we show that widely deployed setups such as Fail2Ban monitoring BIND9 can be defeated with less than 65 Mbps of DNS traffic. Further, we show that when replacing core components of the IPS architecture with their higher-performance equivalent, iptables with eBPF and regex matching with Hyperscan, the logging backends themselves become the bottleneck. Therefore, we present FIPS, a new IPC designed for high-performance logging that bypasses the kernel and reduces copying of the log messages to a minimum. FIPS uses per-thread lock free shared memory ring buffers, supporting multiple independent consumers reading the same log stream at their own pace. FIPS offers both a native API and a drop-in replacement for the syslog interface. Our evaluation with BIND 9 shows that FIPS introduces almost no overhead compared to disabled logging, logs more requests than any other evaluated framework, and enables the IPS to ban malicious clients $2.5\times$ faster than with file logging while sustaining $2^{16}$ attacking clients at one million requests per second.
Read the full abstract

Materials Science

Weisfeiler-Lehman subtree encoding for Bayesian optimization of atomic configurations

The efficiency of Bayesian optimization (BO) of atomic configurations depends strongly on how configurations are encoded. We introduce the Weisfeiler-Lehman (WL) subtree kernel, which views configurations as element-labeled graphs and measures their similarity by how many local structural patterns they share, into Bayesian-optimization-based configuration search. Because this kernel is reproduced as the plain inner product of explicit features (L$^2$-normalized histograms of local topological patterns), introducing it reduces to introducing the corresponding features: the encoding enters existing BO frameworks as an ordinary descriptor. In a benchmark ground-state configuration search of cubic BC$_2$N evaluated with a universal machine-learning interatomic potential, the WL encoding reached the ground state almost immediately after a shared random initialization of 100 samples in every one of five independent rounds (108$\pm$5 evaluations on average), whereas the one-hot baseline required 280$\pm$122 evaluations; the WL-driven sampler first exhausted the degenerate ground-state group and then discovered the metastable degenerate groups from the bottom up, in order of increasing energy.
Read the full abstract

All authors Akira Kusaba, Tatoshi Yonemori, Tetsuji Kuboyama, Yoshihiro Kangawa

Rings and Algebras

The injective envelope of simple modules over Leavitt path algebras II: simples arising from exclusive cycles

Let $K$ be any field, $E$ any directed graph, and $L_K(E)$ the associated Leavitt path algebra. As described first by Chen, and subsequently generalized by Ara and Rangaswamy, for each cycle $c$ in $E$ one can build the simple left $L_K(E)$-module $V_c^E$, and then more generally $V_{p(x),c}^E$ (where $p(x)$ is an irreducible polynomial in $K[x,x^{-1}]$). A cycle $c$ is called exclusive in case none of the vertices of $c$ is the base of any cycle other than $c$. In our main result we provide an explicit description of the injective envelope of $V_c^E$, and then more generally of $V_{p(x),c}^E$, for each exclusive cycle $c$. Our method involves defining an $L_K(E)$-module structure on an appropriately-built $K$-vector space of infinite series. Our main result significantly generalizes previous work of the authors, in that the result holds for all graphs (finite or not), and all exclusive cycles.
Read the full abstract