Peer Revenue Publish or profit

Papers

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

All of arXiv

Friday 4 September

Artificial Intelligence

ProtLingo: Efficient Protein Language Modeling via Conditional Memory and Expert Routing

Proteins perform diverse cellular functions, and even single amino-acid substitutions can alter stability, activity, or molecular interactions. Protein language models (PLMs) provide a scalable approach for modeling such sequence–function relationships from unlabeled sequences, but increasing the size of dense Transformer backbones often brings substantial computational cost without consistently improving mutation-sensitive prediction. We introduce ProtLingo, an efficient PLM framework that augments a pretrained single-sequence backbone with conditional local memory and sparse expert routing. ProtLingo maps contextual residue representations into route-specific discrete codes, composes centered local windows into latent $N$-gram addresses, and retrieves reusable residual signals associated with recurring local sequence contexts. In parallel, selected feed-forward blocks are upcycled into sparse Mixture-of-Experts layers with shared and routed experts, enabling residue-dependent computation while activating only a subset of parameters. Experiments on protein fitness prediction, FLIP benchmarks, and supervised contact prediction show that ProtLingo achieves competitive performance with a 150M-scale backbone, including strong parameter efficiency on mutation-effect prediction and preserved long-range structural representations.
Read the full abstract

All authors Mingrui Li, Sixian Shen, Minzhang Li, Ruiyi Zhang, Kexin Zhang, Jiakai Zhang, Jingyi Yu

Computation and Language

Can Activation Steering Capture Multidimensional Authorship Style?

Activation steering has shown promise for controlling LLM generation along well-defined attributes, but it remains unclear whether it can handle the multidimensional and hard-to-define nature of authorship style. We ask whether structured contrastive prompting along rhetorically-motivated dimensions can construct rich style representations directly in activation space, bypassing the need for natural language style descriptors or dedicated training. We find that the resulting directions share a common authorship backbone while conflicting on aspect-specific residuals that carry genuine stylistic signal, explaining why naive aggregation fails. We operationalize this in Aspect-Aware Activation Steering (A3S), a training-free framework that merges per-aspect contrastive directions with interference-aware aggregation and tunes steering strength per instance. A3S improves authorship style transfer where it is genuinely multi-aspect, outperforms a trained baseline in preference evaluations on out-of-domain benchmarks, and keeps target-exemplar overlap consistently low.
Read the full abstract

Computer Vision

An Attention-Guided Global and Local Fusion Framework for Lesion-Focused Image Classification

Lesion-focused image classification presents a core analytical challenge, as discriminative signals are often sparse, spatially dispersed, and easily obscured by background noise, while conventional convolutional neural networks (CNNs) process entire images uniformly and may dilute signal relevance. This study hypothesizes that adaptive fusion of global contextual information and lesion-focused local information can improve classification performance compared with using either representation independently. We propose a three-branch, attention-guided deep learning framework built on Densely Connected Convolutional Network-121 (DenseNet-121) to improve feature attribution, interpretability, and classification reliability. The architecture consists of a global branch that learns representations from full images, followed by Gradient-weighted Class Activation Mapping (Grad-CAM) to generate attention maps that highlight prediction-relevant regions and produce masked inputs, and a local branch enhanced with a Convolutional Block Attention Module (CBAM) to extract refined spatial and channel-wise features from these focused regions. An adaptive fusion branch integrates global and local representations by learning instance-specific weights, allowing dynamic prioritization between contextual and localized information. The framework is evaluated on a synthetic Spot Pattern Dataset (SSPD) and three benchmark datasets, including skin lesion, guava leaf, and grape leaf image datasets, where the fusion branch outperformed the individual global and local branches, reaching 97.75% accuracy on the skin lesion dataset and 99.64% on the guava leaf dataset. The results highlight the value of attention-guided architectures in healthcare analytics by improving model transparency, strengthening feature relevance, and supporting more reliable data-driven decision-making in medical image analysis.
Read the full abstract

All authors Mst Shafia Tasnima, Md Samaun Elaheea, Tanjim Taharat Aurpab, Md Musfique Anwar

Earth and Planetary Astrophysics

Signatures of rocky debris accretion in AF-Type planet hosts

The gas-giant planet-metallicity correlation plays a fundamental role in our understanding of planet formation. We determine in a homogeneous way the metallicity and individual abundances of a sample of A and early-F stars, with and without known planetary companions, and discuss their abundance distribution and trends. We find that stars with planets generally exhibit higher abundances of both iron-peak and volatile elements compared to stars without known planets although in most cases this tendency is not statistically significant. The $\rm {C/O}$ ratio remains similar between both samples, while the $\rm {Mg/Si}$ ratio is shifted towards higher values in the general planet host sample. When the sample is limited to fully radiative stars (spectral type earlier than F5), the statistical significance of carbon overabundances vanishes. In contrast, elements such as O and Ni remain significantly overabundant. The $\rm {C/O}$ and $\rm {Mg/Si}$ ratios in this radiative subsample are statistically indistinguishable from that of the comparison sample. The persistent overabundance of O and Ni, combined with $\rm {C/O}$ and $\rm {Mg/Si}$ ratios similar to those of stars without planets in fully radiative hosts, points to the ongoing surface pollution by volatile-poor, silicate-rich rocky debris, rather than the accretion of gas giants. These findings might position early-type main-sequence stars as a crucial dynamical missing link in the chemical life-cycle of planetary systems: bridging the initial dust-trapping phase in pre-main sequence transition discs, the dilution of this superficial pollution during the fully convective red giant phase, and the eventual dynamically-driven accretion of asteroid material observed in polluted white dwarfs
Read the full abstract

All authors J. Maldonado, I. Mendigutía, S. Barceló Forteza, G. M. Mirouh, B. Montesinos, E. Villaver

Probability

Height fluctuation for Lozenge Tilings of Polygons

We establish Gaussian free field fluctuations for uniformly random lozenge tilings of simply connected polygonal domains with $3d$ sides whose directions cycle through the three lattice directions. More precisely, assuming that the liquid region is connected and that the boundary data do not force the height at any interior point, we prove that the fluctuations of centered height function converge to the Gaussian free field in the liquid region, confirming a prediction of Kenyon and Okounkov from 2007. We introduce a tiling action function that encodes the geometry of the limit shape through its critical points. The action function has a complex conjugate pair of critical points in the liquid region, repeated real critical points on the arctic boundary, and distinct real critical points in the frozen region. Using this tiling action function, we construct an approximation to the inverse Kasteleyn matrix in terms of explicit single-contour and double-contour integrals and prove that the approximation is uniform throughout the polygonal domain. The convergence to the Gaussian free field then follows from standard kernel computations.
Read the full abstract

Quantum Physics

Shapley Valuation of Finite-Copy Quantum Data Depends on Physical Access

Data valuation asks how learning utility should be attributed to training data contributors. Most classical formulations begin after data have become reusable records, so the physical readout of the data is effectively fixed. Finite-copy quantum data are different: unknown states are consumable physical systems, and the same supplied states and downstream task can yield different Shapley values under different physical access models. Our framework makes this dependence explicit by treating physical access as a component of quantum data valuation itself. We establish an exact connection between physical-access advantage and contributor-level data valuation. For nested access models, we prove that the maximal downstream utility gain enabled by richer physical access exactly determines the largest symmetric Shapley ranking-reversal margin. More generally, for arbitrary access-model pairs, including non-nested ones, we derive an exact geometric characterization of the possible shifts of the full Shapley attribution vector. For fixed learning pipelines, we further obtain an operational Shapley-observable representation for finite-copy valuation. Numerical experiments demonstrate that identical quantum samples can receive different values and rankings when only the physical access model is changed. These results establish that quantum data value is not an intrinsic property of the underlying states alone, but emerges from the interaction between quantum states, physical access, and the downstream learning task.
Read the full abstract

Machine Learning

Learning-Augmented Algorithms: Guarantees, Construction Mechanisms, and System-Level Implications

Learning-augmented algorithms use fallible predictions while retaining formal performance guarantees. This survey synthesizes prediction interfaces, error measures, consistency–robustness trade-offs, and five representative construction mechanisms across online optimization, caching, learned data structures, graph problems, and mechanism design. An orthogonal theorem-level axis distinguishes achieved upper bounds from matched asymptotic dependence. Formal guarantees are separated from empirical systems evidence, with explicit treatment of prediction cost, feedback, and composition. The resulting synthesis states sufficient conditions for limited end-to-end reasoning and delineates open problems in cost-aware prediction, endogenous error, semantic predictors, and benchmarking.
Read the full abstract

All authors Hailiang Zhao, Xueyan Tang, Peng Chen, Zenong Ye, Tianjun Zhu, Jianwei Yin, Shuiguang Deng

Cryptography and Security

Injected and Leaked: Actively Inducing Side-Channel Leakage Using Electromagnetic Injection and Hardware Nonlinearity

Electromagnetic (EM) side-channel leakage and injection are typically treated as distinct physical phenomena, threatening data confidentiality and integrity respectively. This work investigates how EM injection can be used to amplify side-channel leakage that is otherwise infeasible. We introduce a novel framework for Injection-Induced EM Side Channels to enable integrated, closed-loop EM security analysis. Our theoretical modeling and experimental measurements reveal that nonlinear hardware components, such as ubiquitous amplifiers, analog-to-digital converters, and power converters, can modulate secret electrical signals onto an injected EM carrier and thus upconvert low-frequency secrets into measurable EM emissions. By tuning the injection frequency and amplitude, adversaries gain the ability to actively shape the effective spectrum and entropy of the resulting leakage. We design InjectEave attack and demonstrate eavesdropping on the audio played through wired and wireless headphones from up to 30 m away with accessible RF equipment, as well as in through-wall scenarios, and characterize injection-induced EM leakage of other low-frequency secrets such as power consumption of smart home devices and analog sensor inputs. Case studies further demonstrate how the proposed techniques enable closed-loop eavesdropping and manipulation of landline-phone conversations. Finally, we analyze the broader security challenges and mitigations.
Read the full abstract

All authors Haoran Yan, Ziyu Shao, Shuhao Zhang, Qinhong Jiang, Yan Long

Computer Vision

CLON: Cue-Calibrated Linguistic Object Onboarding for Zero-Shot 6D Pose Front-Ends

Novel-object 6D pose pipelines depend on front-end hypotheses that retain target instances under clutter and occlusion without overwhelming downstream pose estimation. We introduce Cue-Calibrated Linguistic Object Onboarding (CLON), a training-free proposal-and-matching front-end. CLON converts multi-view CAD renderings into compact linguistic memories that provide object-specific conditioning to a frozen proposal generator. Constructed offline and reused across scenes, these memories augment generic prompts before segmentation; subsequent template verification assigns object identity. As a complementary onboarding step, CLON derives fixed cue weights from template statistics for semantic, appearance, and geometric verification. On seven BOP-Classic-Core datasets, CLON improves detection AP by 8.2 percentage points (pp), segmentation AP by 6.2 pp, and downstream pose AR by up to 4.1 pp over the evaluated CNOS and SAM-6D front-ends. In a real-world localization-to-grasp study under clutter and occlusion with 100 trials per method, CLON's improved localization doubles grasp success relative to the strongest evaluated baseline. Controlled ablations evaluate linguistic conditioning, multi-view construction, and verification. Analysis of a smaller candidate set with higher class-agnostic recall on five datasets.
Read the full abstract

Hardware Architecture

A Piecewise-Linear Approximation-based Energy-Efficient Error-Optimized Unsigned Square Rooter for Accuracy-Critical Applications

Approximate computing improves energy efficiency in error-resilient applications, but square root units remain challenging due to the trade-off between hardware cost and computational accuracy. This paper presents an energy-efficient, error-optimized, piecewise-linear approximation-based unsigned square rooter (EOSQR) for 2n-bit inputs that achieves high accuracy with low hardware complexity, using only simple arithmetic and shift operations. The EOSQR design is implemented in Verilog-HDL and evaluated on a 16-bit benchmark synthesized on an Artix-7 FPGA. Compared to representative state-of-the-art approximate square rooters, EOSQR achieves the lowest error among accuracy-critical designs while delivering 61.91 percent resource savings, 77.54 percent power savings, and 53.11 percent latency reduction relative to a precise restoring array-based square rooter. To enable holistic evaluation, a Composite Efficiency Metric (CEM) is introduced to jointly capture accuracy and energy efficiency. EOSQR is further validated across representative image-processing workloads, including Sobel edge detection, K-means colour quantization, and K-nearest-neighbour (KNN) classification. Experimental results demonstrate that EOSQR achieves high computational accuracy with a superior CEM-based accuracy-hardware efficiency trade-off while maintaining visual quality and classification performance, making it well-suited for real-time edge-embedded systems.
Read the full abstract

Artificial Intelligence

DODR: Deterministic Operator-Driven Reasoning in Latent Space

Autoregressive (AR) large language models formulate reasoning as token-level probabilistic sampling, which induces three fundamental defects in complex logical reasoning: error accumulation, probability substituting necessity, and the linear-chain information bottleneck. This paper proposes the Deterministic Operator-Driven Reasoning in Latent Space architecture (DODR), which reconstructs reasoning as reasoning-graph computation in a high-dimensional linear-algebraic space. Reasoning states are represented as snapshot vectors whose primitives are semantic units (phrases or sentences) rather than tokens, and each inference step is a deterministic matrix operation with no token sampling. Peirce's three inference types are formalized as three trainable matrix operators: a rank-deficient deduction operator (information collapse), a full-rank induction operator (information expansion), and an abduction operator defined as the Moore-Penrose pseudo-inverse of deduction (information hypothesizing). We prove that the operator set is minimal and complete given Peirce's trichotomy, that no single "super-operator" can realize all three types (a rank obstruction), and that reasoning graphs are Turing-complete with contractive backflow converging by Banach's fixed-point theorem. Experiments on 503 sample records (420 deduplicated samples) across dedicated and end-to-end settings show: deduction loss converges to 1.40e-05; induction achieves 0.9996 generalization coverage with 20/20 hard vetoes on counterexamples; abduction solutions exceed the random baseline by 28x with judgment accuracies of 72.5% (58/80, Wilson 95% CI [61.9%, 81.1%]) and 81.7% (49/60, CI [70.1%, 89.4%]); frozen operators attain 100% (60/60) on unseen cross-domain deduction. The architecture provides a structural zero-hallucination guarantee and a three-layer continual-learning mechanism. All data and code are released.
Read the full abstract

Computer Vision

CoMLP: Cooperatively-Gated MLPs for Fine-Grained Cross-Modal Information Fusion in Medical Image Segmentation

Multi-modal medical images and clinical reports provide complementary anatomical, functional, and semantic information for medical image segmentation. Effectively exploiting these heterogeneous sources requires fine-grained cross-modal information fusion that preserves subtle spatial details while capturing semantic dependencies across modalities. Existing fusion approaches frequently rely on cross-attention, whose computational burden increases rapidly with spatial resolution, making dense cross-modal interaction difficult on high-resolution feature maps, particularly for volumetric medical images. In this work, we propose CoMLP, a cooperatively-gated MLP module for fine-grained cross-modal information fusion in medical image segmentation. CoMLP models cross-modal dependencies through cooperative cross-gating, built upon complementary regional and dilated MLP interactions, to capture local and global cross-modal dependencies. We further develop a multi-source fusion architecture in which CoMLP performs both inter-image fusion across imaging modalities and vision-language fusion between visual features and textual reports, enabling heterogeneous information to be integrated without relying on dense cross-attention. Extensive experiments on five medical segmentation benchmarks, covering 2D/3D images, clinical reports, multiple imaging modalities, and diverse anatomical regions, demonstrate consistent improvements over state-of-the-art multi-modal and language-guided segmentation methods. Ablation studies further show that fine-grained interaction at high spatial resolutions and complementary local-global fusion are critical to the performance gains. These results demonstrate the potential of MLP-based interaction as an effective alternative for fine-grained cross-modal information fusion in medical image segmentation.
Read the full abstract

All authors Mingyuan Meng, Shuchang Ye, Mingjian Li, Zhenyu Zhao, Jinman Kim, Lei Bi

Analysis of PDEs

On Liouville problems for $(p,q)$-Laplacian inequalities on Finsler measure spaces

We study the Liouville property for nonnegative weak solutions of the $(p,q)$-Laplacian elliptic differential inequality $$\Delta^{m}_{p}u(x)+\Delta^{m}_{q}u(x)+V(x)u^{s}(x)\leq 0$$ and the associated parabolic differential inequality $$\partial_t u(x,t) \geq \Delta^{m}_{p}u(x,t)+\Delta^{m}_{q}u(x,t)+V(x,t)u^{s}(x,t)$$ on a forward geodesically complete noncompact Finsler measure space $(M,F,m)$ with finite reversibility. Under several sets of integral growth conditions on the positive potential function over certain annular domains, we prove that any nonnegative weak solution vanishes almost everywhere. The proofs of our results are essentially based on the nonlinear capacity method.
Read the full abstract

Machine Learning

Dynamic Heterogeneous Graph Representation Learning: A Survey

Graph representation learning (GRL) serves as a canonical paradigm for modeling complex networks. However, real-world AI systems inherently manifest as evolving heterogeneous entities with complex interactions, posing significant challenges to static or homogeneous modeling. To address these complexities, representation learning for Dynamic Heterogeneous Graphs (DHGs) has emerged as a vital approach for learning low-dimensional representations that simultaneously preserve structural semantics and temporal dynamics. This survey presents the first systematic review of DHG representation learning methods. We first introduce a unified formal definition that encompasses both discrete-time and continuous-time DHGs from the perspective of temporal granularity. Building upon this formulation, we propose a novel algorithm-centric taxonomy that categorizes existing literature, including early embedding-based approaches, graph neural network (GNN)-based models, and relatively recent Transformer-based DHG methods, while explicitly highlighting their intrinsic modeling biases with respect to dynamic granularity. Furthermore, we summarize representative applications of DHG representation learning, along with commonly used datasets and benchmarks. Finally, we discuss promising research directions that guide future advances in this rapidly evolving field.
Read the full abstract

All authors Huan Liu, Pengfei Jiao, Jie Yin, Hongjiang Chen, Zhidong Zhao

Artificial Intelligence

Diffusion Language Models for Mobile Edge Agentic AI: Foundations, Applications, and Challenges

Diffusion language models (DLMs) offer a non-autoregressive alternative for mobile edge agentic artificial intelligence (AI) by refining tokens through iterative denoising rather than left-to-right decoding. Compared with autoregressive Transformer-based large language models (LLMs), DLMs can update multiple uncertain tokens in parallel and exploit bidirectional context throughout the generation process, enabling more flexible quality-latency trade-offs beyond fixed sequential decoding. These properties are particularly attractive for edge agents, where partial refinement, early exit, and constraint-guided correction can reduce response delay and communication overhead while improving robustness under noisy, incomplete, or dynamic contexts. This survey reviews DLM foundations and analyzes their suitability for edge settings under latency, memory, energy, bandwidth, privacy, and reliability constraints. We cover resource-efficient architectures, training and inference acceleration, compression, edge/cloud deployment, communication-aware serving, Internet of Things (IoT)/wireless applications, and evaluation of DLM-based agents. We further discuss open issues in long-context state management, split inference, trustworthy execution, multimodal grounding, and reproducible benchmarking. The goal is to connect DLM modeling properties, including bidirectionality, parallel refinement, controllability, and quality-latency elasticity, with system-level requirements of future mobile edge intelligence.
Read the full abstract

All authors Chenqi Li, Minghui Min, Dusit Niyato, Wei Ni

Computers and Society

On Being Prepared: Automated Vehicle Incident Management Exercise Practices

Incident management (IM) has evolved over recent decades to cover an ever-expanding array of hazards and systems. Barring real-life experience, exercises are a key tool in developing an effective IM program. As part of enterprise resilience and operational readiness, IM practitioners design exercises to understand and build capabilities for efficiently and effectively responding to incidents. Combining research and policy on an emerging transportation technology, automated vehicles (AVs), with established practices for IM, we describe what makes exercises effective and how they can be used to identify gaps and develop capacity. A range of exercise types exist, from workshops, to tabletops, to drills, and to full-scale exercises. Every stage of an exercise - preparing, setting up, facilitating, closing, and assessing - is interconnected and should further the exercise's objectives. AV IM practitioners can maximize the utility of a mature exercise program through exercise quality, diversity, and volume. The need to respond to AV incidents, from collisions to natural hazards, is inevitable. This paper tailors existing practices to a series of frameworks, predicated on the unique socio-technical nature of AVs to provide first-of-its-kind guidance for AV IM exercises. Additionally, these frameworks are company-neutral to enable intercompany and inter-organization collaboration.
Read the full abstract

All authors Laura Fraade-Blanar, Jackson Graves, Aaron Clark-Ginsberg, Sam Cooper, Jeremy Kaufman, Jackie Lindsey, Piet van Os

High Energy Astrophysical Phenomena

The angular structure of the GW170817 jet from prompt emission alone

We determine the angular structure of the GW170817 jet by the prompt emission alone, without afterglow fitting, circumburst density or microphysical parameters. We assume that GRB 090510 and GW170817 have outflows of the same kind, observed respectively on-axis and at the interferometric viewing angle of $20^\circ$. We support this assumption with independent gravitational-wave data showing compatible binary masses and radiated energies. We show that for an observer whose beaming cone is filled with outflow, the point-source Doppler scalings do not apply: $E_{\rm iso}=4\pi\epsilon(\theta_v)$, with $\epsilon$ the energy radiated per unit solid angle along the line of sight, while the peak energy follows $E_{\rm p,i}\propto\Gamma(\theta_v)$. We obtain $n={\rm d}\ln\epsilon/{\rm d}\ln\Gamma = 3.76\pm0.29$ from the ratio of the two bursts, with no free parameter and no assumed angle. This excludes four structures in common use at $4.7\sigma$ to $13\sigma$; three remain above $4\sigma$ across the full reported range of the peak energy of the GW170817 jet. Two prompt spectra fix no angular scale; supplying it with the core Lorentz factor of GRB 090510 and the viewing angle gives $\epsilon\propto\theta^{-7.4}$ outside a core of $2^\circ$–$5^\circ$, in agreement with the width inferred from $367$ short bursts, the outflow remaining relativistic at $\Gamma=33$ on the line of sight. The exponent exceeds what the Lorentz boost of a uniform comoving flow can produce, so the structure is intrinsic to the outflow and not a consequence of the boost. The same structure fixes the emission radius, which contributes $0.41$~s of the $1.74$~s delay between the gravitational-wave signal and the gamma-rays, the remainder being the launch and breakout of the jet, with no free parameters. We conclude that the faintness lies in the structure of the GW170817 jet, not in the de-beaming of a bright core.
Read the full abstract

Computer Vision

LUMIN: Lifecycle-Utility-aware and Metadata-Informed Network for Visual Anomaly Detection

Industrial anomaly detection faces two engineering bottlenecks: memory bank construction latency and inference efficiency. Traditional sampling algorithms (Farthest Point Sampling, K-Means, etc.) rely on numerous backbone forward passes and iterative distance computations, with construction times ranging from minutes to hours; heavy computation components such as multi-scale feature extraction struggle to meet the millisecond-level real-time requirements of production lines. This paper focuses on sampling efficiency and inference optimization for industrial deployment with two core contributions: (1) PSP (Plugin Sampler Pipeline)—a four-stage adaptive memory bank sampling pipeline based on 18-dimensional pixel metadata and five complementary visual plugins. PSP completes all sampling with zero backbone forward passes; coarse filtering is sub-second numerical sorting, and metadata extraction is a one-time offline cost. PSP supports progressive deployment and incremental updates. (2) Two engineering optimization strategies—parallel memory bank similarity computation (reducing inference memory and latency by over 95%) and stratified pixel sampling for large-scale evaluation (reducing computation time by 20$\times$ while keeping metrics stable). As a vehicle for validation, we introduce LUMIN (Lightweight Universal Manufacturing Inspection Network) with extreme segmentation-head compression, systematically exploring the accuracy-efficiency frontier against strong baselines. Experiments on five benchmarks demonstrate that PSP matches state-of-the-art sampling accuracy at near-random construction cost (341$\times$ faster than FPS), while inference optimizations reduce evaluation time by 20$\times$ with negligible accuracy loss.
Read the full abstract

Operating Systems

Adaptive Context Parallelism for Production LLM Serving

As LLM context windows expand and input sequences grow longer, serving systems face increasing computational and memory demands. Context parallelism (CP), which partitions the input sequence across multiple ranks to parallelize the computation, has therefore become increasingly important for efficient LLM serving. However, existing CP-enabled systems either rely on static CP configurations or adjust the CP degree only for active requests or batches. In this paper, we present Vertumnus, an adaptive CP serving system designed for heterogeneous and evolving workloads. At the request level, Vertumnus routes requests among workers with different CP degrees using a placement cost that combines predicted queuing delay, cache-aware prefill time, and GPU-time cost. At the cluster level, Vertumnus adapts the worker composition through seconds-scale split and merge operations as workload demand changes. Vertumnus further introduces a global prefix-cache management policy that coordinates cache placement and replication among workers with the same or different CP degrees, preserving cache locality as request assignments and worker composition change. Experiments on a 64-GPU cluster with public and production workloads show that, under the highest evaluated loads, Vertumnus reduces mean TTFT by up to 28.1% and improves token-weighted SLO attainment by up to 13.3 percentage points over the strongest baseline.
Read the full abstract

All authors Jiarui Guo, Rongle Wang, Peijun Huang, Zongwei Lv, Ziqing Wang, Kan Liu, Tao Lan, Lin Qu, Xiaolin Wang, Tong Yang

Machine Learning

Persistent Teacher Anchoring for Tool-Using Agents

Distillation is common in LLM post-training, where on-policy knowledge distillation (OPKD) uses student-generated trajectories to prepare the student for downstream RL. At each state, the student matches a next-token distribution supplied by the teacher. As the rollout enters states the teacher would not visit, the teacher-student distribution gap can accumulate. In tool use, this gap becomes consequential because student-written calls execute before supervision and their observations shape later prefixes. Proposer-verifier generation addresses this drift by letting the teacher decide which student-proposed text is retained during generation. Existing formulations govern text but leave tool execution outside their scope. We propose Persistent Teacher Anchoring (PTA), a student-induced but teacher-committed rollout construction. PTA retains chunk-level verification and adds turn-level commitment, allowing a call to reach the environment only after the teacher has verified the entire turn. Treating verified chunks as atomic generation units, we introduce persistent lookahead, which fills idle rollout capacity by advancing future samples and carrying unfinished ones across student updates under the fixed verifier. Across Search-R1-style retrieval and DeepEyes-style perception RL, applying PTA before downstream RL improves macro best@4 by 2.5 and 2.8 points over OPKD under the same downstream RL budget, while lookahead improves throughput by 24%.
Read the full abstract

All authors Hyun Bin Park, Kyungho Song, Sangmin Lee, Du-Seong Chang

Machine Learning

A Robust Watermark-based Fingerprint Framework for GNNs Ownership Verification

The high training cost of Graph Neural Networks (GNNs) has raised growing concerns regarding model ownership infringement, such as model stealing and unauthorized misuse. To verify model ownership and prevent significant economic losses, two groups of GNN Ownership Verification (OV) methods have been proposed: watermark-based methods and fingerprint-based methods. However, these methods typically face three limitations: (1) the performance degradation of protected models caused by out-of-distribution (OOD) watermark graphs with respect to the training set; (2) the unrealistic assumption that surrogate models have been trained on a watermark-containing training set; and (3) over-reliance on specific output levels for fingerprint extraction. In this paper, we propose a Robust watErMArk-based fingeRprint frameworK for GNNs, named REMARK. REMARK first generates carefully crafted in-distribution watermark graphs that maximize output differences between GNN models, thus mitigating OOD-induced performance degradation. REMARK then extracts robust fingerprints from these output differences to verify GNN ownership, thereby removing the assumptions that surrogate models must be trained on a watermark-containing dataset or expose specific output levels. Extensive experiments across widely used real-world datasets and GNN architectures demonstrate that REMARK achieves state-of-the-art OV accuracy and robustness while preserving the utility of protected models.
Read the full abstract

All authors Han Zhang, Yan Wang, Guanfeng Liu, Pengfei Ding, Huaxiong Wang, Kwok-Yan Lam

Combinatorics

Albertson's Conjecture for Chromatic Numbers at Most 29

Albertson's conjecture asserts that every finite simple graph $G$ with $\chi(G) \ge r$ satisfies $\operatorname{cr}(G) \ge \operatorname{cr}(K_r)$. Building on Cranston's verification for $r \le 24$ and his reduction of $r \in \{25,26\}$ to three residual orders, we eliminate those residual cases and then prove the cases $r=27,28,29$. The first structural ingredient is a Kempe-chain construction: if a $k$-critical graph has a vertex of degree $k-1$, then it contains a branch-clean essential immersion of $K_k$. Essential immersions are crossing-monotone, so a critical counterexample must have minimum degree at least $k$. For $r=27$, this one-unit degree gain, Gallai's join structure, critical-graph edge bounds, and induced-subgraph averaging close every possible order. For $r=28$ and $r=29$, the remaining near-$2r$ orders are converted to dense complements. Stehlík's coloring theorem makes the odd-order complements factor-critical; a clique-partition obstruction yields an anti-tight matching property; and Tutte barriers, Hall-type expansion, and deficit bookkeeping eliminate the final cases. At order 58 for $r=29$, Rabern's coloring inequality handles the regular case, while the last degree-deficit-two case is reduced to two disjoint triangles and a finite barrier analysis.
Read the full abstract

Robotics

Continuous Cognitive Coverage for Autonomous Robots via Event-Dependent Cognitive Treatment and Learning

Autonomous robots continuously encounter objects, changes, and situations, and every event admitted into cognition should receive an appropriate cognitive treatment rather than remain untreated until an explicit task requires attention. However, existing task-driven, reactive, or fixed-reasoning approaches generally process only selected events or apply predefined reasoning procedures, making it difficult to provide continuous cognitive coverage with differentiated treatment. This paper proposes a continuous cognitive coverage framework in which every cognitively admitted event is assigned an event-dependent cognitive treatment according to its state, context, and history. Different events may therefore invoke description, memory, risk prediction, planning, diagnosis, analogy, or other learned treatments. Familiar events can be processed automatically by learned mechanisms, whereas unfamiliar or uncertain events invoke explicit deliberation or fallback reasoning. Multiple cognitive processes can be suspended, resumed, and interleaved so that cognitive processing continues as new events arrive or existing events await evidence. Validated experiences are continuously learned to automate, refine, and revise event-specific treatments. Experiments achieve 96.76% structured treatment accuracy with 93.66% automatic processing, 92.64% cognitive coverage under bursty-delayed workloads, and 79.53% continual-learning joint accuracy, with novel-event reuse reaching 100% automatic processing.
Read the full abstract

Cryptography and Security

Memory-Efficient Designs for Word-Wise Universal Fully Homomorphic Encryption

Fully Homomorphic Encryption (FHE) enables computation on encrypted data, preserving privacy throughout analysis. While its privacy is very strong, FHE is much slower to execute than the original computation. In particular, due to the recent success in accelerating its compute, the performance bottleneck shifts to the memory, especially considering that FHE magnifies the data size by orders of magnitude, resulting in a low arithmetic intensity. We propose BXT, an FHE optimization framework that mitigates the memory bottleneck through four techniques: (1) ciphertext compression, which regenerates ciphertext components from seeds during execution; (2) ciphertext serialization, which packs coefficients as bit arrays and unpacks them during L2-to-L1 transfer; (3) delayed seed generation, which defers PRNG-heavy offline work across aggregated operations; and (4) ciphertext digit pruning guided by fault-aware training tailored for Universal FHE. On CNN inference, the BXT-CSO50 configuration effectively achieves up to 3.8$\times$ speedup over the 100x GPU baseline with less than 1% accuracy loss at 50% comparison precision.
Read the full abstract

All authors Ardhi Wiratama Baskara Yudha, Erwin Eko Wahyudi, Rian Adam Rajagede, Qian Lou, Yan Solihin

Combinatorics

Spectral properties of the resistance Laplacian with applications to data clustering and anomaly detection

The resistance Laplacian is a graph matrix associated with the effective resistance metric and provides a global counterpart of the classical graph Laplacian. Although it inherits several fundamental properties of the ordinary Laplacian, including a connected graph partitioning theorem analogous to that of Fiedler, its intrinsic spectral structure has remained largely unexplored. In this paper, we develop a structural theory of the resistance Laplacian. We derive a canonical decomposition that separates its intrinsic, average, and deviation components, thereby revealing how the global geometry induced by effective resistance differs from the local geometry encoded by the ordinary Laplacian. Building upon this decomposition, we establish several structural and spectral properties of the associated deviation operator, obtain variational characterizations of the largest eigenvalue and its corresponding eigenspace, and express the resistance Laplacian in Laplacian coordinates, thereby elucidating the relationship between the eigenspaces of the two operators. Finally, we formulate resistance-based graph partitioning objectives whose spectral relaxations recover the dominant eigenvector of the resistance Laplacian, providing a variational interpretation of the connected partition theorem. Experimental results on synthetic and real-life datasets demonstrate the effectiveness of the proposed framework for graph partitioning, data clustering, and exploratory anomaly detection.
Read the full abstract

Artificial Intelligence

Shadow Queries for Private Retrieval in Vector Databases

Large language models (LLMs) increasingly rely on information retrieval (IR) systems, such as Retrieval-Augmented Generation (RAG), to incorporate domain-specific knowledge without costly re-training. These systems often store pre-computed document embeddings in cloud-based vector databases. However, such embeddings are vulnerable to embedding inversion attacks (EIAs), which can reconstruct their underlying text. Existing defenses, such as adding noise or scaling embeddings, often provide limited privacy or significantly reduce retrieval utility. We propose SHAQ (shadow query generation), a semantic-decomposition and embedding-decoupling defense against EIAs. SHAQ is based on the insight that EIAs rely on the strong coupling between an embedding and its original text. Instead of storing document embeddings directly, SHAQ uses a generative language model to create diverse shadow queries that capture different semantic aspects of each document. These queries are then encoded and stored in place of the original document embeddings, thereby decomposing document semantics and decoupling stored embeddings from the source text. Experiments across diverse IR datasets show that SHAQ substantially improves privacy while preserving retrieval utility, achieving a recovery rate as low as 0.2104, defending up to 19.50% more tokens than baseline defenses, and reaching up to 0.7967 MAP@10 with up to 5.53% utility improvement. These results demonstrate that semantic decomposition and embedding decoupling provide an effective alternative to directly modifying embeddings for defending against EIAs.
Read the full abstract

All authors Xinguo Feng, Zhongkui Ma, Zihan Wang, Chuan Yan, Guowei Yang, Alsharif Abuadbba, Guangdong Bai

Computers and Society

When Does an Interpretation Count as Established? The Formation, Evaluation, and Responsibility of Interpretation in Generative AI

Generative AI research has increasingly evaluated factuality, citation, coverage, and report structure. Yet passing such local checks does not by itself show that a humanistic interpretation has been established. This paper asks how an interpretation comes to be recognized within sociotechnical processes. It introduces three connected concepts. Interpretive appearance names the gap between the finished form of an output and the publicly traceable process through which materials, counterevidence, and revisions constrained the judgment. The evaluation contract names the bounded materials, tasks, criteria, permitted inferences, and failure conditions within which a local judgment is valid. Standing substitution names the unwarranted conversion of a genuine local pass into a stronger claim that an interpretation, result, or research capability has been established, without commensurate new evidence or bridging arguments. The paper then examines responsibility for judgment: a text may acquire recognition while no public structure remains for stating reasons, answering objections, revising, downgrading, or withdrawing the conclusion. Humanistic scholarship provides a revealing test because new materials and conceptual distinctions can alter both the question and the criteria of evaluation. The paper therefore develops delayed closure as a practice of keeping recognized interpretations revisable and proposes five public requirements concerning materials and versions, evidential roles, failure, contract revision, and responsibility. The argument is conceptual and normative: it does not claim to offer a benchmark or to determine whether models possess understanding. It instead explains why local evaluation, finished textual form, and public recognition must not be treated as sufficient evidence that an interpretation has been formed.
Read the full abstract

High Energy Astrophysical Phenomena

Through a glass, darkly: a combined framework for estimating fast radio burst host galaxy and population properties in an era of uncertain host identification

The identification of fast radio burst (FRB) host galaxies, and subsequently their redshifts ($z$), has allowed the FRB dispersion measure (DM) to be used to probe the cosmological distribution of ionised gas, and study the properties of the FRB population itself. However, current methods cannot account for FRBs with uncertain host galaxy associations, leading to underutilisation of data, and potential biases towards nearby, bright hosts. In this work, we develop a methodology which can. We do so by combining three ingredients - the zDM code, for modelling the Macquart relation; PATH, for statistical host galaxy identification; and a set of models able to describe the intrinsic FRB host galaxy distribution - into a single formalism. We prove the fidelity of our formalism by using a synthetic set of FRB observations, with simulated hosts sampled from galaxy catalogues, and show that it reproduces intrinsic host galaxy properties even for FRB localisation uncertainties of 30", where a traditional PATH analysis produces no confidant host associations. When applied to a sample of FRBs localised by the Australian Square Kilometre Array Pathfinder, we confirm previous results showing that FRBs prefer host galaxies fainter than that predicted by star-formation, consistent with an exponential surface density scaling as $0.41^{+0.13}_{-0.09}$ times the half-light radius. We encourage the application of this formalism to data-sets from other FRB-hunting instruments.
Read the full abstract

All authors C.W. James, B.C. Andersen, L. Marnoch, J.L. Hoffmann, N. Loudas, J.X. Prochaska, S.D. Ryder, M. Woodland

Number Theory

On multiply induced characters

We investigate finite group structures that yield complex irreducible characters induced simultaneously from two or more distinct subgroups. When such a group is realized as the Galois group of a Galois extension of number fields, this phenomenon implies that multiple subfields share the same Artin $L$-function.
Read the full abstract