arXiv subjects

Yue Wu

Publications and source records attributed to Yue Wu.

At least 19 recordsLinked to original sources

arXiv2026

RHINO: Guided Reasoning for Mapping Network Logs to Adversarial Tactics and Techniques with Large Language Models

Modern Network Intrusion Detection Systems (NIDS) generate vast volumes of low-level alerts that remain semantically fragmented, requiring labor-intensive manual correlation with high-level adversarial behaviors. Existing automation, namely rule-based systems and machine learning classifiers, has critical limitations: rule-based approaches fail to adapt to novel attack variations, while machine learning methods lack contextual awareness and treat tactic-technique (TT) mapping as syntactic matching rather than reasoning. Although Large Language Models (LLMs) show promise in cybersecurity, preliminary experiments reveal that existing LLM-based methods frequently hallucinate technique-tactic pairs or produce decontextualized mappings due to single-step classification. We introduce RHINO, a framework that decomposes LLM-based attack analysis into three structured phases: (1) behavioral abstraction, translating raw logs into contextualized narratives; (2) partitioned inference over the ATT&CK tactic space, generating candidate techniques by evaluating behavioral evidence against MITRE ATT&CK knowledge; and (3) knowledge-grounded refinement, cross-referencing predictions with official MITRE definitions to reduce hallucinations. We evaluate RHINO on three benchmarks (DAPT2020, CICIDS2017, IoT23) across four backbone models (ChatGPT-4o, Gemini 2.5 Flash, Claude Sonnet 4, and DeepSeek V3). Under a scenario-documented single-label formulation covering 11 distinct techniques, RHINO achieves an average top-1 accuracy of 87.34% under micro-averaging and 63.77% under macro-averaging, surpassing the strongest baseline by 16.78 and 13.60 points respectively. These results show that structured LLM reasoning can effectively translate low-level network observations into high-level adversarial behaviors.

cs.CR↗

arXiv2026

Select-TTT: Query-Conditioned Selective Fast Weight Updates for In-Place Test-Time Training

Test-time training (TTT) enables a deployed language model to internalize incoming context on-the-fly into a small set of fast-weight parameters, rather than merely treating it as transient context for attention. In-Place TTT recently made this idea practical for large language models by repurposing the down-projection of standard MLP blocks as fast weights. However, we find that its benefit largely disappears on realistic web long-context question answering. We attribute this to In-Place TTT's uniform write schedule: every chunk is written with equal strength, so irrelevant content dilutes query-relevant information in fast-weight memory. Building on this insight, we propose Select-TTT, a model-agnostic, query-conditioned rule for In-Place TTT that scales each chunk's write strength by its relevance to the user query. This relevance signal requires no supervision and no new parameters: it is read directly from the model's frozen output head, as the probability mass that a chunk's hidden states assign to the query tokens. A temperature-scaled softmax turns these scores into learning-rate multipliers, leaving the chunk-wise, context-parallel computation of In-Place TTT intact. Select-TTT consistently improves long-context question answering on two benchmarks and two backbones. These findings highlight the value of query-conditioned fast weight writes for practical test-time training on long-context question answering, which is also potentially promising for long-horizon agent tasks involving multiple user queries over evolving contexts.

cs.LG↗

arXiv2026

Planar energetics do not determine twin persistence in nanocrystalline aluminium

Planar energetics are the standard descriptors of twinnability in face-centred cubic metals and the usual validation targets for interatomic potentials. Here we show that they do not determine whether deformation twins persist in nanocrystalline aluminium. Two machine-learned potentials reproducing the same zero-load slip curves and closely spaced layered energy surfaces predict, from a common separated-fault precursor under identical loading, retained twin coverage of 64\% and 14\%. The divergence traces to forces within distorted partial cores acting along the trailing translation that would remove an outer fault. Planar energetics do set the margin: an outer fault more than halves the incremental barrier for middle-plane shear, yet twin completion and fault removal remain separated by only 10--21 mJ m\textsuperscript{-2}. Whether an assembled twin persists is decided instead by the orientation of local traction relative to the full-slip bisector and by the arrival order of partial-slip fronts from the grain-boundary network.

cond-mat.mtrl-sci↗

arXiv2026

EP251129a: A Fast X-ray Transient with Two X-Ray Plateaus and Its Off-axis Afterglow

Extragalactic fast X-ray transients (EFXTs) detected by the Einstein Probe are extending the parameter space of relativistic stellar explosions toward softer and longer-lived transients. Here we report EP251129a/GRB 251129A at z=0.505, which exhibits two X-ray plateaus: an early plateau lasting $\sim 10^{4}\,\rm s$ followed by a rapid decay with spectral softening, and a second plateau extending to 60 days, placing it at the extreme long-duration end of Gamma Ray Burst (GRB) plateaus. The early optical emission evolves independently of the X-ray, indicating a different origin; whereas the unusual radio afterglow shows a rapid rise of $F_ν\propto t^{1.9-2.5}$ and a delayed peak at $\sim36-40$ days, providing strong evidence for an off-axis geometry. We find that the early X-ray behavior resembles those of ultra-long GRBs (ULGRBs) and can be naturally explained by the internal dissipation from a newborn millisecond magnetar undergoing gravitational-wave-dominated spin-down before collapsing into a black hole. The early optical, radio and late X-ray emission are well described by an off-axis two-component jet afterglow with additional energy injection, possibly powered by fallback accretion onto black hole. EP251129a therefore connects the emerging EFXT population with ULGRB-like central-engine activity and off-axis afterglow, offering a rare opportunity to probe both the evolution of central engine and jet structure in a single event.

astro-ph.HE↗

arXiv2026

Retention-constrained fine-tuning of a pre-trained aluminium MACE potential for non-equilibrium defect dynamics and plasticity

Adapting a pre-trained machine-learning interatomic potential for non-equilibrium defect dynamics and plasticity requires improving defect accuracy without degrading the near-equilibrium elastic and vibrational response. Here we achieve this balance for the aluminium MACE potential QP17 by integrating conditional readout uncertainty, iterative full-parameter updates and protocol-compatible relative-energy anchoring. Sensitivities over 144 linear readout weights guide acquisition, with within-batch covariance updates and evolving neural representations. After four active-learning rounds (32 acquired configurations and 400 gradient updates), force-uncertainty acquisition reduces mean development force error from 22.43 to 21.52 meV/Angstrom, outperforming random sampling (22.70), geometry-diversity selection (22.46) and uniform training across all 104 candidate labels (22.71) under the same update budget. All three force-uncertainty runs satisfy bulk-mechanical retention criteria: historical relative-energy error decreases from 0.375 to 0.207 meV/atom, while harmonic phonon root-mean-square drift remains below 0.05 THz. Matched objective ablations reveal that relative-energy anchoring suppresses harmonic drift. Out-of-sample evaluations across planar faults, twin boundaries and grain boundaries demonstrate transferable gains in defect energies and deviatoric stresses, even though virial stresses are never fitted. A closed-loop molecular dynamics and density functional theory (MD-DFT) pilot confirms that automated acceptance gating intercepts updates that violate historical elastic-group energy or vibrational tolerances. Together, these components establish a retention-constrained fine-tuning protocol for improving plastic-deformation observables while preserving the tested bulk mechanical response.

cond-mat.mtrl-sci↗

arXiv2026

Energy stability of explicit--implicit--null time discretizations for nonlinear degenerate parabolic equations with state-dependent coefficients

We study nonlinear degenerate parabolic equations with state-dependent coefficients admitting a scalar energy potential. The explicit--implicit--null (EIN) method avoids severe time step restrictions of explicit time stepping and costly Newton--Raphson iterations required by a fully implicit treatment of the nonlinearity. Using orthogonal projection of the Kirchhoff transform, we prove nonlinear energy stability for first- to third-order EIN time stepping coupled with a wide range of spatial discretizations that admit self-adjoint positive semidefinite discrete operators. Under an upper bound for the diffusion coefficient on relevant stage segments, two IMEX coefficient conditions yield sufficient stabilization bounds without any time step restriction. The estimates dissipate the original convex energy and allow degenerate diffusion. We also extend the analysis to diffusion systems with a scalar energy potential in a fixed positive definite metric, and to consistent positive-weight quadrature discretizations. Applications include porous-medium flow, nonlinear heat transfer in plasma, and multicomponent diffusion.

math.NA↗

arXiv2026

Timed Recovery Safe Gradient Flows for Constrained Nonlinear Programming

This paper studies how to relax pointwise feasibility in continuous-time dynamics for inequality-constrained nonlinear programming while retaining a quantitative feasible-recovery guarantee. Existing safe gradient flows enforce forward invariance of the feasible set, which yields an anytime-feasible dynamics but can be overly restrictive for optimization because it suppresses transient infeasible motion that may be essential for effective descent. To address this limitation, we propose a timed recovery safe gradient flow that replaces pointwise feasibility with fixed-horizon feasible recovery. The proposed construction augments the state with a timer that records the elapsed infeasible duration and computes a minimum-norm correction to the nominal gradient flow through a pointwise convex quadratic program. We show that the resulting augmented safe set is forward invariant and that every trajectory initialized from the feasible set returns to it within every time window of length $τ$. We further show that, in a strict interior inactive regime, the closed-loop dynamics reduces exactly to the nominal gradient flow and recovers the corresponding local convergence behavior. Numerical examples illustrate how the recovery horizon shapes the closed-loop behavior and corroborate the theoretical analysis.

math.OC↗

arXiv2026

VibeTraining-Bench: Data-Trainer-Harness Co-Design for Iterative System Improvement

Recursive self-improvement (RSI) requires agents to diagnose system bottlenecks and coordinate improvements across training and inference. We refer to the agents' capability to jointly optimize training data, the trainer, and the inference harness as cross-component co-design. We introduce VibeTraining-Bench, a benchmark for evaluating this capability across tasks spanning terminal operation, software engineering, multi-turn tool use, and mathematical reasoning. Within each task, research agents start from an identical initial agent system under a fixed resource budget. Guided by validation feedback, they iteratively refine data, trainers, and inference harnesses, then select and submit a final agent system for evaluation on the test set. To prevent reward hacking, we enforce three safeguards: test data isolation, separate research and grading containers, and network restrictions for research agents. Our evaluation results show that frontier research agents substantially improve target agent systems. For example, on SWE-bench Verified, GPT-6-Astra raises the target agent system's performance from 32.9% to 53.7%, comparable to the human-developed instruct model's 55.4%. However, there is still room to improve their ability to co-design and iteratively refine data, trainer, and harness to achieve cumulative performance gains. We hope that VibeTraining-Bench will serve as a unified platform for advancing RSI, deepening our understanding of cross-component optimization for agent systems. Code and data are available at https://github.com/D2I-ai/VibeTraining-Bench.

cs.AI↗

arXiv2026

Critical functions and asymptotics for fractional Hartree Brezis--Nirenberg problems in low dimensions

We study the Brezis--Nirenberg problem for the fractional Laplacian with an upper critical Hartree nonlinearity on a bounded domain $Ω\subset\mathbb R^N$, in the low-dimensional range $2s<N<4s$. For a continuous potential $a$, we prove that the following properties are equivalent: the Robin function of $(-Δ)^s+a$ is negative somewhere in $Ω$; the associated Hartree--Sobolev quotient ${\mathcal S_{H,s}}(a)$ lies strictly below the best constant ${\mathcal S_{H,s}}$; and ${\mathcal S_{H,s}}(a)$ is attained. This leads to the notion of a critical potential, for which ${\mathcal S_{H,s}}(a)={\mathcal S_{H,s}}$ and every nontrivial decrease of $a$ lowers the infimum. For a critical potential $a$ and $V\in L^\infty(Ω)$, we determine the leading term in the asymptotics of ${\mathcal S_{H,s}}(a+\varepsilon V)$ as $\varepsilon\to0^+$ and describe the blow-up of almost-minimizers, including the concentration speed and the location of the concentration point. Compared with the Sobolev case, the nonlocal nonlinearity produces several new interaction terms; a key ingredient is a proof of non-attainment at the threshold based on the sharp Hardy--Littlewood--Sobolev inequality on the sphere.

math.AP↗

arXiv2026

Cowork-Gym: Synthesizing Interactive Work Environments at Scale

Cowork agents navigate dependencies across tools and files while adapting to changing user requirements and edits to shared workspaces. Scaling such environments requires maintaining consistency among task objectives, executable resources, user behavior, and success criteria. We introduce Cowork-Gym, a framework that synthesizes Cowork environments through a build-execute-revise loop that uses execution evidence and review feedback to validate and repair generated environments. To capture realistic Cowork dynamics, a stateful user simulator responds to the agent's actions through dialogue and direct edits to the shared workspace. We synthesize 64.9K environments across 14 application scenarios and use them to post-train 9B, 27B, and 35B-A3B Qwen models with supervised fine-tuning, reinforcement learning, and on-policy distillation. Averaged across six agentic benchmarks, these models outperform their corresponding base checkpoints by 8.7, 6.4, and 4.1 points, respectively. Ours-27B ranks second overall among the evaluated models and achieves the highest average scores on PawBench and ClawEval, outperforming models with over 100B total parameters on both benchmarks.

cs.AI↗

arXiv2026

Non-isotopic surfaces in $T^4\#(S^2\times S^2)$: an example

We prove that there exist infinitely many embedded tori with a common geometric dual in $T^4\#(S^2\times S^2)$ that are homotopic, diffeomorphic, but not isotopic to each other, even after arbitrary many external stabilizations. These surfaces are obtained by applying the Norman trick to a fixed immersed surface, using non-homotopic tubing arcs. The isotopy classes of these surfaces are distinguished by homotopy classes of the 2-handles (relative to the boundary) in the complement of the image of the $0$- and $1$-handles.

math.GT↗

arXiv2026

From Impact to Insight: Dynamics-Aware Proprioceptive Terrain Sensing on Granular Media

Robots that traverse natural terrain must interpret contact forces generated under highly dynamic conditions. However, most terrain characterization approaches rely on quasi-static assumptions that neglect velocity- and acceleration-dependent effects arising during impact and rapid stance transitions. In this work, we investigate granular terrain interaction during high-speed hopping and develop a physics-based framework for dynamic terrain characterization using proprioceptive sensing alone. Through controlled hopping experiments with systematically varied impact speed and leg compliance, our measurements reveal that quasi-static based assumptions lead to large discrepancies in granular terrain property estimation during high-speed hopping, particularly upon touchdown and controller-induced stiffness transitions. Velocity-dependent drag alone cannot explain these discrepancies. Instead, acceleration-dependent added-mass effects - associated with grain entrainment beneath the foot - dominate transient force responses. We integrate this force decomposition with a momentum-observer-based estimator that compensates for rigid-body inertia and gravity, and introduce an acceleration-aware weighted regression to account for increased force variance during high-acceleration events. Together, these methods enable consistent recovery of granular resistance across locomotion conditions, closely matching linear-actuator ground truth. Our results demonstrate that accurate terrain inference during high-speed locomotion requires explicit treatment of acceleration-dependent granular effects, and provide a foundation for robots to characterize complex deformable terrain during dynamic exploration of terrestrial and planetary environments.

cs.RO↗

arXiv2026

Specialized machine learning force fields for materials dynamics

Machine learning interatomic potentials (MLIPs) are transforming atomistic simulations by accessing unprecedented length and time scales. While pretrained equivariant graph neural networks achieve robust zero-shot performance for near-equilibrium properties across broad chemical spaces, their translation to complex materials dynamics remains fundamentally challenged by out-of-distribution reactive states, representation biases, and computational scaling limits. In this Review, we examine how physics-driven specialization extends the applicability of MLIPs to complex dynamical systems. We systematically evaluate the structural trade-offs in MLIP design: the undersampling of highly strained configurations, the prohibitive computational overhead of high-order message-passing architectures, and the necessity of nonlocal interactions for open and field-coupled systems. Through four demanding application contexts - electrified interfaces, compositionally fluctuating open systems, multiphase evolution, and large-scale fracture - we establish a framework for observable-specific validation. Highlighting the complementary roles of universal foundation models and task-specific potentials, we emphasize that targeted adaptations must be rigorously benchmarked against intended observables. We conclude with a roadmap for developing physically consistent, hardware-aware force fields that seamlessly connect electronic-structure accuracy to macroscopic materials phenomena.

cond-mat.mtrl-sci↗

arXiv2026

Closed-loop domain generation enhances battery health diagnosis

Reliable electrode-level degradation diagnosis is essential for accurate health estimation and non-destructive monitoring in advanced battery management. Diagnosis models developed under controlled laboratory conditions can fail when operating conditions change, while acquiring sufficient full-lifespan data in new target domains remains prohibitively time-consuming and practically unfeasible. Here, we propose a closed-loop domain generation framework to enable cross-domain battery health diagnosis under dynamic conditions. Physics-constrained domain generation is achieved through Variational Autoencoder-based latent extrapolation, which maps the source data and few-shot early-life target domain data into a shared latent space and generates cross-domain samples within physical boundaries. Bayesian optimization closes the generation-diagnosis loop by automatically calibrating the sampling, latent regularization, extrapolation, perturbation, and generation-volume hyperparameters using leakage-free development validation, thereby strictly excluding later-life target data from optimization and ensuring the quality and quantity of generated data. Numerical results demonstrate that, across four cross-domain cases, the proposed framework achieves average root mean square errors of 1.53%, 2.15%, and 2.02% for positive electrode active material loss, negative electrode active material loss, and lithium inventory loss, respectively, whereas traditional methods suffer from notably higher errors or even complete failure. This framework provides a general pathway for onboard and physics-consistent electrode-level degradation diagnosis, independent of downstream diagnosis models.

eess.SY↗

arXiv2026

Prism: Dynamic Sparse Attention for Native 2K Joint Video-Audio Generation Model Training

Natively training joint video-audio generation models at higher resolutions empowers them to learn richer visual details and sharper motion dynamics. However, full attention incurs quadratic cost and, as resolution increases, spreads attention over increasingly redundant tokens, diluting learning signals for informative content and disrupting pretrained priors. Existing sparse attention methods either target training-free acceleration or overlook the unique structure of joint video-audio data, where cross-modal interactions are inherently concentrated around sound-producing regions. To address this, we propose Prism, a dynamic sparse attention framework for natively training joint video-audio generation models at 2K. In particular, Prism organizes the token sequence into spatiotemporal macro-zones, enabling the attention structure to adapt to local content. For each zone, it estimates local information structure via video feature variance along the channel and feature norms from the audio-to-video cross-attention, jointly capturing how visual content varies directionally and how strongly audio influences each visual region. Based on these signals, Prism dynamically assigns a tailored block shape to each zone, applying finer partitioning along axes of rapid visual content variation and strong audio-visual coupling. This encourages tokens within each block to remain semantically coherent, allowing block-level features to capture both visual content and joint video-audio interaction patterns. Prism further adopts a hybrid block selection strategy to dynamically determine per-query sparsity. Experiments show that Prism achieves 2.5$\times$ training speedup compared to full attention, while surpassing it in generation quality.

cs.CV↗

arXiv2026

Predicting properties of Scrooge ensembles with high accuracy and low sample complexity

The Scrooge ensemble associated with a density matrix $ρ$ is the maximally random ensemble whose average density matrix is $ρ$. It has attracted growing interest for its applications to quantum many-body dynamics and quantum information. The intrinsic statistical properties of the Scrooge ensembles are encoded in their higher-order moments, yet they are nontrivial to analyze and prepare for they feature a non-polynomial Haar integrand, blocking direct use of Haar integration machinery. Existing physical mechanisms to sample multiple copies of states from the Scrooge ensembles, meanwhile, typically require costly postselection, rendering them inefficient for large system size. In this work, we develop efficient and accurate methods for analyzing and implementing Scrooge moments. We first introduce a polynomial approximation to the Scrooge $k$-th moments whose error is tunable down to a threshold that improves from inverse polynomial to exponential in $1/||ρ||_\infty$ over previous results. Building on our approximation, we then give the first efficient quantum algorithm that estimates observable expectation values using $\widetilde{\mathcal{O}}(k^3/ε^5)$ copies of $ρ$ for any target error $ε$ above this threshold. We further provide a construction of a block encoding of the Scrooge $k$-th moment when the preparation circuit of $ρ$ is known. Of independent interest could be a key subroutine in our algorithms, which efficiently implements projection onto the $k$-fold symmetric subspace for $ρ^{\otimes k}$ using $\mathcal{O}(k^3/ε^2)$ samples of $ρ$, avoiding the naive $k!$ postselection cost of such a projection. These results provide new tools for both the theoretical analysis and practical prediction of properties of Scrooge ensembles, with potential applications to quantum many-body dynamics and quantum information processing.

quant-ph↗

arXiv2026

Hierarchical Reasoning Model

Reasoning, the process of devising and executing complex goal-oriented action sequences, remains a critical challenge in AI. Current large language models (LLMs) primarily employ Chain-of-Thought (CoT) techniques, which suffer from brittle task decomposition, extensive data requirements, and high latency. Inspired by the hierarchical and multi-timescale processing in the human brain, we propose the Hierarchical Reasoning Model (HRM), a novel recurrent architecture that attains significant computational depth while maintaining both training stability and efficiency. HRM executes sequential reasoning tasks in a single forward pass without explicit supervision of the intermediate process, through two interdependent recurrent modules: a high-level module responsible for slow, abstract planning, and a low-level module handling rapid, detailed computations. With only 27 million parameters, HRM achieves exceptional performance on complex reasoning tasks using only 1000 training samples. The model operates without pre-training or CoT data, yet achieves nearly perfect performance on challenging tasks including complex Sudoku puzzles and optimal path finding in large mazes. Furthermore, HRM outperforms much larger models with significantly longer context windows on the Abstraction and Reasoning Corpus (ARC), a key benchmark for measuring artificial general intelligence capabilities. These results underscore HRM's potential as a transformative advancement toward universal computation and general-purpose reasoning systems.

cs.AI↗

arXiv2026

MemGuard: Preventing Memory Contamination in Long-Term Memory-Augmented Large Language Models

Memory-augmented large language models extend reasoning beyond a fixed context window by maintaining long-term memory across interactions. However, existing memory systems often collapse stable user facts, episodic events, and behavioral rules into a shared space, allowing functionally distinct memories to be retrieved and used as interchangeable evidence. We identify this failure mode as heterogeneous memory contamination, where context-specific events become overgeneralized claims, or semantically relevant but functionally incompatible memories mislead generation. To this end, we introduce MemGuard, a type-aware memory framework that preserves functional memory boundaries during memory construction and retrieval. It assigns each memory an explicit functional role at write time, maintains relations across type-isolated memories, and selectively composes evidence only from necessary memory types, reducing contamination from irrelevant or functionally incompatible evidence. Across hallucination and long-horizon conversation benchmarks, MemGuard improves memory reliability by up to 28.27% while retrieving up to 5.8x fewer memory tokens than prior methods. These results suggest that reliable long-term reasoning depends on principled organization and selective use of heterogeneous memory.

cs.CL↗