arXiv subjects

Xiao Liu

Publications and source records attributed to Xiao Liu.

At least 37 records · Page 2Linked to original sources

arXiv2026

Seeing Semantic Shift: Difference-Aware Sentence-Level Temporal Segmentation of Sign Language Videos

Recent advances in sign language understanding have achieved impressive success on short, single-sentence videos, yet their performance drops sharply when applied to long, continuous sign language videos. To bridge this gap, we focus on a challenging and realistic setting: Visual-only Sentence-level Sign Language Segmentation (Vis-SSLS), which aims to partition continuous sign language videos into non-overlapping sentence-level segments without any caption assistance, serving as a crucial prerequisite for downstream recognition and translation tasks. However, sentence transitions in sign language are often smooth and visually ambiguous, lacking explicit pauses or posture resets. As a result, static frame representations may fail to capture the subtle temporal changes that indicate sentence boundaries. To address this challenge, we propose \textbf{SignShift}, a difference-aware segmentation framework that explicitly models frame-to-frame feature variation as semantic cues for sentence boundary detection. First, to model the feature variation, we design a Temporal Difference Module, which incorporates full-frame, facial, and hand cues, and employs inter-frame differencing to learn multi-scale temporal variations that capture both fine-grained local kinematics and global semantic transitions. Second, to mitigate over- and under-segmentation issues, we design a Segment Count Prediction module, which predicts the number of sentences to guide boundary selection. Extensive experiments on benchmark datasets demonstrate that SignShift substantially outperforms existing methods, validating its effectiveness.

cs.CV↗

arXiv2026

MDRC: A Deployable State-Recovery Defense for Traffic Signal Control under Sensor Corruption

Traffic Signal Control (TSC) is a safety-critical cyber-physical system that relies on real-time sensing. Corrupted observations caused by adversarial perturbations or sensor failures can propagate from the sensing layer into the controller and degrade traffic efficiency. Existing robust Reinforcement Learning (RL)-based TSC methods often suffer from limited cross-city generalization, high inference latency, and weak recovery under partial observability. We present MDRC (Meta-Diffusion-based framework for Resilient traffic signal Control against adversarial attacks and sensor failures), a post-detection state-recovery defense inserted between sensing and control. MDRC reconstructs trustworthy traffic states before they are consumed by the controller. It combines Denoising Diffusion Implicit Models (DDIM) for efficient state recovery with Reptile meta-learning for a transferable initialization across cities. We provide an optimization-based view of the DDIM recovery dynamics and establish a recovery-error bound that separates score approximation, numerical discretization, and initialization mismatch. Across seven real-world-derived CityFlow benchmarks, MDRC reduces Average Travel Time by 6.77% under stochastic and policy-aware attacks and by 12.75% under structured sensor loss, while improving state-recovery fidelity. We further evaluate 3,600 seconds of real roadside measurements with 50% of detector channels disabled and integrate MDRC into a hardware-in-the-loop traffic-signal stack. Over a 9.16-hour run with 32,389 sensing/control cycles, the system achieves 99.79% decision availability, produces no out-of-plan recommendations, and requires approximately 38 ms of component-wise processing per one-second control interval.

cs.CR↗

arXiv2026

What Was Once Learned May Need to Be Unlearned: Machine Unlearning for Deprecated API Knowledge in Large Language Models

Large language models (LLMs) for code completion may generate deprecated APIs because their pre-training corpora contain code from historical library versions. Existing approaches use inference-time intervention, model editing, or machine unlearning, but multiple plausible completions make predefined replacements restrictive. Moreover, existing studies rarely verify whether models exhibit the targeted deprecated behavior or evaluate unintended changes to other APIs. We conduct a systematic empirical study of machine unlearning for deprecated API knowledge and construct MUDAPIBench, a behavior-grounded benchmark with over 7,000 model-specific instances derived from 145 deprecated-to-up-to-date API mappings across eight Python libraries. Instances are retained only when the original model generates the target deprecated API. We evaluate eight representative unlearning methods across three code LLMs on deprecated API forgetting, up-to-date API generation, preservation of other and unrelated API behaviors, general code-generation capability, and efficiency. Results show that Gradient Difference (GD) provides the best overall trade-off, suppressing deprecated APIs while preserving other capabilities with moderate computational costs. Further analyses reveal substantial variation across libraries and show that APIs deprecated after the model's training-data cutoff are harder to forget. Layer-wise analyses indicate that GD achieves effective forgetting with comparatively controlled internal changes.

cs.SE↗

arXiv2026

Explainable Recommendations at Scale: LLM Rationales for YouTube Music Artist Discovery

Modern music streaming platforms face a persistent tradeoff: exploiting familiar content versus driving the exploration of novel items. While users frequently desire discovery, they hesitate to select unknown artists over proven favorites. Providing transparent, natural language rationales that explain why an unexplored item is recommended lowers this barrier. However, while Large Language Models (LLMs) excel at this nuanced explainability, their real-time deployment is severely bottlenecked by prohibitive inference costs and computational overhead. In this paper, we present an industry case study of a decoupled recommendation architecture that successfully scales exploration without compromising latency. Our system isolates LLM inference asynchronously offline, pre-computing personalized candidate pools of undiscovered artists alongside tailored rationales. Large-scale online A/B experiments validate our design. We demonstrate that combining LLM-backed recommendations with these explanatory rationales significantly reduces the trust barrier for new content, yielding statistically significant improvements in both user exploration and overall engagement on the discovery surfaces.

cs.AI↗

arXiv2026

GRAVA: Grounded Reasoning-to-Action Representation and Learning for Autonomous Driving

Driving vision-language-action (VLA) models increasingly reason before acting, but their intermediate reasoning is often weakly grounded in physical scene evidence and loosely connected to executable behavior. We present GRAVA, a framework built around Grounded Reasoning-to-Action (GRA), which unifies grounding, reasoning, and action generation in a single autoregressive stream. GRA links action-relevant language references to 2D visual regions and ego-centric physical states, organizes object interactions and decisions in a trajectory-anchored typed graph, and serializes this structure into grounded reasoning. A single VLM generates this reasoning followed by a compact Executable Planner action that is deterministically decoded into a continuous trajectory. We further introduce an agentic GRA data construction pipeline that combines forward scene grounding with backward trajectory anchoring, and use it to build GR-NavSim with 2.2M grounded question-answer pairs and 70K GRA reasoning traces. A progressive training strategy develops grounded cognition through pre-training, establishes the reasoning-to-action interface through imitation, and improves driving behavior through reinforcement learning and exploration. Using about 60% of the available human driving demonstrations for action supervision, GRAVA-8B achieves state-of-the-art performance among purely autoregressive driving models on the full NAVSIM benchmark. On an internal long-tail benchmark, full GRA improves key-object compliance and Closed-loop Driving Score by 19.3% and 20.5% over action-only prediction, respectively. These results show the benefit of preserving action-relevant physical evidence from grounded reasoning through executable action generation.

cs.CV↗

arXiv2026

Nonequilibrium Apparent-Horizon Thermodynamics in Effective Loop Quantum Cosmology

In this work, we investigate the thermodynamic consistency of the apparent horizon in effective loop quantum cosmology (LQC). By retaining the standard definitions of horizon thermodynamic quantities, we find that the modified LQC dynamics leads to a departure from the classical equilibrium thermodynamic description. This deviation can be naturally described by a nonequilibrium formulation characterized by an internal entropy production term. The resulting entropy production consistently recovers the classical and de Sitter limits. These results suggest that the apparent horizon in effective LQC admits a coherent nonequilibrium thermodynamic interpretation.

gr-qc↗

arXiv2026

Expert-Space Exploration in MoE Reinforcement Learning

Reinforcement learning (RL) has become central to post-training of large language models. Recent advances in RL for Mixture-of-Experts (MoE) models have primarily focused on improving optimization stability and training efficiency, while treating the expert selection as a fixed component. Since routing determines the sparse computation paths that induce output distributions, expert selection offers an additional source of rollout diversity. Through empirical analysis, we find that perturbing expert routing effectively alters model output and increases rollout diversity, which is similar to increasing the decoding temperature. However, direct perturbation can activate unsuitable experts and substantially degrade rollout quality. Motivated by these observations, we introduce Expert-Space Exploration Reinforcement Learning (ESRL), an architecture-aware framework that explicitly explores the expert-routing space of MoE models. ESRL preserves high-confidence experts as anchors, and restricts stochastic routing to a plausible candidate pool, thereby retaining reliable computation paths. The perturbation strength is further adapted according to router entropy to avoid over-perturbation. To mitigate the routing mismatch introduced by perturbation, ESRL records the expert paths used during rollout and replays them during policy optimization. Experiments demonstrate that ESRL achieves the best performance across MoE backbones with top-K, top-1, and shared-expert routing, as well as across mathematics, science, and code tasks without additional sampling or computational cost. Specifically, ESRL on Qwen3-30B-A3B achieves the best among all compared methods, improving average Pass@1 and Pass@8 over GRPO by 3.2 and 4.5 percentage points, respectively. Further analyses of expert utilization and training dynamics provide insights into how exploiting MoE-specific routing structure benefits RL training.

cs.CL↗

arXiv2026

MSign: An Optimizer Preventing Training Instability in Large Language Models via Stable Rank Restoration

Training instability remains a critical challenge in large language model (LLM) pretraining, often manifesting as sudden gradient explosions that waste significant computational resources. We study training failures in a 5M-parameter NanoGPT model scaled via $μ$P, identifying two key phenomena preceding collapse: (1) rapid decline in weight matrix stable rank (ratio of squared Frobenius norm to squared spectral norm), and (2) increasing alignment between adjacent layer Jacobians. We prove theoretically that these two conditions jointly cause exponential gradient norm growth with network depth. To break this instability mechanism, we propose MSign, a new optimizer that periodically applies matrix sign operations to restore stable rank. Experiments on models from 5M to 3B parameters demonstrate that MSign effectively prevents training failures with a computational overhead of less than 7.0%.

cs.LG↗

arXiv2026

STEP: State-Aware Task Estimation and Planning with Multi-Modal LLMs for Human-Robot Collaboration

Effective human-robot collaboration in industrial settings requires robots to understand human intentions and assist with task planning, reducing workload. Recent works have explored the use of Multi-modal Large Language Models (MM-LLMs) for task planning in such data-scarce scenarios, leveraging in-context learning to interpret user actions and generate long-horizon action plans in natural language. However, MM-LLMs inherently lack an understanding of system states and do not track state transitions, often leading to hallucinated actions that deviate from the intended goal. Additionally, generating action plans in natural language tends to limit the generated plans to a high level, introducing ambiguity in action execution. To address these limitations, we propose the State-aware Task Estimator and Planner (STEP), which prompts a MM-LLM to explicitly estimate the state of the system and predict the state transitions resulting from executed actions. By forecasting future states alongside actions, STEP ensures task-convergent planning while also providing additional assistance parameters necessary for executing the predicted actions. We evaluate STEP in a simulated environment using a robot assembly task. Our approach outperforms the state-of-the-art by 32.8% in action executability and 14.8% in final-state error.

cs.RO↗

arXiv2026

SANE: State Anomaly Neutralization for Stable Extreme-Context Delta-Rule Models

Delta-Rule recurrent models maintain a fixed-size state, enabling $O(1)$ inference memory but potentially becoming unstable under extreme-context extrapolation. By tracking RWKV-7 over sequences of up to 100M tokens, we empirically identify a distinct failure pattern: \textbf{localized norm explosion atop a relatively sparse substrate}, rather than global state saturation. Analysis of the recurrent update suggests that persistent decay keeps weakly updated entries small, whereas uneven injections allow a few channels to accumulate extreme values. Motivated by this diagnosis, we propose \textbf{State Anomaly Neutralization (SANE)}, which applies adaptive $\tanh$ compression at chunk boundaries while preserving the intra-chunk parallel structure. Within a safe threshold range ($3 \le α\le 5$), SANE matches the baseline on 11 short-context reasoning benchmarks with no statistically significant degradation. After a 100M-token prefix, which exceeds the training length by over $24{,}000\times$, SANE retains functional reasoning ($33.46$--$35.56$) while the baseline encounters numerical overflow. In contrast, overly permissive thresholds ($α\ge 8$) remain numerically stable but lose reasoning capability entirely, showing that numerical stabilization alone does not guarantee functional reasoning and revealing a capacity--stability trade-off in state compression.

cs.LG↗

arXiv2026

Efficient, precise DFT calculations of NMR shieldings: Revisiting the finite field approach

Absolute nuclear magnetic shielding constants and relative chemical shifts are second-derivative response properties that underpin the interpretation of Nuclear Magnetic Resonance (NMR) spectra and provide critical insights into the structural and electronic environments of diverse chemical systems. However, their accurate computation is often constrained by the need for method-specific, analytical response implementation, and therefore is particularly challenging for non-variational, correlated wavefunction methods where analytical second derivatives are frequently unavailable. Here we present a scalable framework that computes NMR shielding tensors through finite magnetic field differentiation of complex-valued, gauge-including atomic-orbital (GIAO) based self-consistent field (SCF) calculations. By combining hybrid MPI and OpenMP parallelization and resolution-of-identity (RI) approximation-based Coulomb (J) and Exchange (K) implementation and mixed numerical/analytical derivatives computational scheme, we achieve very good efficiency for hybrid density functional theory (DFT) NMR shielding, relative to existing state-of-the-art analytic implementations across realistic chemical systems of various sizes. Forward differentiation with optimal step sizes derived from rigorous error analysis retains favorable numerical errors much smaller than experimental uncertainty or intrinsic DFT errors. The results demonstrate that RI-accelerated finite magnetic field calculations can obtain DFT-level NMR shielding constants very precisely, providing a scalable foundation for extensions to more advanced quantum chemistry methods in future work.

physics.chem-ph↗

arXiv2026

Data-Driven Modeling of Spatiotemporal Dynamics Using Multimodal Imaging Data

Understanding how biological systems evolve across space and time remains a fundamental challenge, particularly when dynamic processes vary substantially across individuals. We present a personalized graph-based dynamical modeling framework for characterizing spatiotemporal biological dynamics from longitudinal multimodal imaging data. The framework constructs individualized brain graphs from MRI and PET measurements and learns patient-specific dynamical parameters governing regional structural and molecular changes. Applied to 1,891 participants from the Alzheimer's Disease Neuroimaging Initiative, the model captures the coordinated evolution of amyloid-$β$, tau, neurodegeneration, and cognition and accurately predicts their future trajectories, outperforming established clinical and neuroimaging benchmarks. Patient-specific dynamical parameters reveal distinct patterns of biological progression and provide improved prediction of future cognitive decline compared with standard biomarkers. Sensitivity analysis further identifies regional network features associated with the propagation of pathological and structural changes, recovering known temporolimbic and frontal vulnerability patterns. These results demonstrate how data-driven dynamical modeling can integrate multimodal longitudinal measurements to uncover individualized spatiotemporal patterns and latent mechanisms of biological change. The framework provides a quantitative approach for studying complex biological dynamics across heterogeneous individuals and establishes a foundation for personalized modeling of progressive biological processes.

q-bio.NC↗

arXiv2026

Displacement-field-driven reconstruction of low energy transport in few-layer PtSe2

In layered semiconductors, a perpendicular displacement field generates an interlayer potential difference that competes with interlayer hybridization, modifying both the band gap and the finite-density electronic states that carry current. Resolving this interplay requires a material lying close to the semiconductor-to-semimetal transition, where moderate electric fields can strongly reshape the low-energy electronic structure. Here, we investigate displacement-field-driven transport in dual-gated semiconducting PtSe2, whose pronounced thickness-dependent electronic structure provides access to this low-band-gap regime. Unlike thinner layers, the displacement-field response is strong in six-layer PtSe2, which lies at the verge of the semiconductor-to-semimetal crossover with only a small residual transport gap. Even weak displacement fields rapidly suppress this residual gap near charge neutrality, driving the system toward a band-overlap regime. At the same time, the conductivity decreases in the heavily hole-doped regime, demonstrating that the displacement field modifies not only the gap but also the conducting valence-band states. Fixed-relaxation-time Wannier transport calculations reproduce both responses, showing that they originate from field-induced band overlap together with reconstruction of the valence-band dispersion. These results establish finite-density transport as a sensitive probe of displacement-field-driven electronic structure reconstruction and extend electrical control beyond conventional band-gap engineering.

cond-mat.mtrl-sci↗

arXiv2026

Markets, Not Planners: Decentralized Orchestration of LLM Agents with Private Information

As LLM agents proliferate, built by different parties and with different capabilities and costs, orchestrating them is more like assembling labor across the economy than a computer calling a subroutine. Existing orchestration is typically centralized, with a single planner assigning every task, but this creates a bottleneck as agent pools grow, requires private information (e.g., agents' execution costs), and can easily be manipulated, such that a single inserted preference nearly doubles a favored agent's task share under a centralized LLM allocator. We introduce AgentLance, a repeated labor market in which agents bid on tasks using their private costs and self-maintained strategy notes, an allocator selects winners from bids and public reputation records, and a VCG-style payment rule rewards cost-aware bidding. Complex tasks are handled by hierarchical delegation: winning agents can decompose work and subcontract it through the same mechanism. Across mathematical reasoning, code generation, knowledge-intensive QA, and agentic tasks, AgentLance matches agents to their specializations, shifts work toward cheaper agents as cost sensitivity rises, and consistently outperforms single-model, centralized-orchestration, and market baselines. Diagnosing market failures, including inaccurate cost self-estimation and sub-optimal bidding, then correcting them in controlled experiments yields further gains, charting a path toward more efficient agent economies.

cs.MA↗

arXiv2026

StageWAM: Joint-Embedding Stage Prediction for World-Action Models in Robot Manipulation

Generalist robot policies aim to map multimodal observations and linguistic task instructions to actions across diverse tasks. However, existing methods typically represent the future as a fixed, short video-action chunk. This short-term future captures local scene evolution for action execution, but it does not explicitly describe the stage-level future that specifies how a task should progress from its current stage to the next. We therefore distinguish two complementary futures for robot manipulation: a short-term physical future to capture local scene evolution and a stage-level semantic future to represent task progress. We introduce StageWAM, which augments a Motus-based World Action Model (WAM) with Stage-JEPA, a goal-conditioned Joint-Embedding Predictive Architecture (JEPA) predictor. Given the current observation and task instruction, Stage-JEPA uses a frozen V-JEPA2 encoder to extract the current-state representation and predicts the latent target of the next inferred stage. Across 50 RoboTwin 2.0 tasks in clean and randomized environments, StageWAM achieves 90.25% overall success and reduces the mean number of execution steps in successful rollouts by 5.97% relative to the strongest baseline.

cs.RO↗

arXiv2026

G0.5: One Autoregressive Stream for Robot Reasoning and Action

The prevailing recipe for Vision-Language-Action (VLA) models couples a pretrained VLM with a separately trained flow-matching action expert. This makes the VLM a context encoder rather than a decision-maker. We introduce G0.5, a pretrained autoregressive VLA in which a single transformer decoder emits reasoning and action tokens under a single objective. Three components make this tractable at foundation-model scale: a learnable cross-embodiment action tokenizer that maps heterogeneous robot actions into a shared vocabulary; a native chain-of-thought stream interleaving task decomposition, object grounding, and action hints with action tokens; and a visual memory module that injects multi-second history through the vision encoder. Because reasoning and action share a single set of weights, the pretrained VLM's capabilities carry over to physical behavior: the model follows instructions closely, and prompts directly steer action granularity, task horizon, and out-of-distribution scene handling without further training. Pretrained on a large collection of robot datasets together with VQA samples, G0.5 surpasses state-of-the-art models across 7 independent regimes: real-world fine-tuning on R1lite and R1pro robots (76.7\% vs.\ 53.3\% for $π_{0.5}$ and 24.4\% for GR00T-N1.7), the 2025 BEHAVIOR Challenge on 50 long-horizon household mobile manipulation tasks using a generalist policy (31.4\% vs.\ 26.3\% for $π_{0.5}$ and 26.1\% for the challenge winner), DROID post-training followed by zero-shot transfer to an unseen environment and objects (82.5\%), a language-following Pick-and-Place benchmark, LIBERO (98.9\%), RoboTwin 2.0 (93.3\%), and SimplerEnv-Bridge (87.3\%).

cs.RO↗

arXiv2026

TideRL: Boosting Agentic RL Goodput with Readiness-Aware Scheduling

Reinforcement learning (RL) for large language models is moving toward multi-turn agentic workloads, where rollout tasks repeatedly pause for external environments, resume with growing contexts, and finish at highly variable times. In this setting, RL training goodput, measured by training throughput, matters more than raw GPU occupancy: GPU waiting and repeated prefill recomputation are pure overhead. We present TideRL, a readiness-aware elastic RL system with Continuous Task Batching, Resource-Aware Ref-Actor Pipelining, and Elastic Resource Scaling. CTB preserves useful rollout state, $\textrm{RA}^2\textrm{P}$ selects between decoupled streaming and colocated aggregation from the ready backlog and arrival interval, and ERS moves ranks between rollout and training using the same readiness signals. Across text-only and multi-modal agentic workloads, TideRL improves RL training goodput by up to 5.6$\times$ over synchronous baselines and over 33% over asynchronous baselines, while reaching similar task performance. It also improves KV cache hit rate by 1.58$\times$, reduces per-step training time by up to 44.3%, and cuts total waiting time by up to 77.6%.

cs.LG↗

arXiv2026

Contemporary AI lacks the imagination to diverge or negate in science

Bold claims that AI will accelerate scientific discovery have raced ahead of evidence from working scientists, yet large-scale, scientist-in-the-loop evidence is scarce. Here we mount the largest evaluation to date, inviting authors of 121,640 recent preprints in biology, medicine, chemistry, and social science to judge large language model (LLM)-generated ideas derived from their own papers. 6,749 representative scientists returned 25,139 rating sets on novelty, feasibility, probability of being true, and favorability of adoption. Three patterns emerge. First, non-reasoning LLMs collapse into a narrow "hivemind" of similar ideas while reasoning models explore a wider hypothesis space, but no model spontaneously proposes null hypotheses, a move humans make more freely. Second, scientists reward ideas resembling their own and prize probability over novelty, though social scientists tolerate risk more than life scientists; senior social scientists are the harshest critics, and their skepticism is earned, as LLMs falter most in pluralistic fields demanding context-aware interpretation and evolving theories. Third, automated evaluators, including LLM-as-a-judge and state-of-the-art (SOTA) models, agree weakly with expert judgment. Retrieval augmentation and scientist persona prompting yield marginal gains. A Qwen3-14B reward model we post-trained on human ratings captures nuances of taste, beats SOTA models by up to 27%, and closes the gap to the consistency of human peer reviewers. An analysis of 39 million papers from 2010 to 2025 links survey findings to macro-level patterns: following ChatGPT's release, null claims are sharply suppressed and ideas contract. Agent-based simulations further suggest that saturated fields should especially prize human uniqueness. For all the hype, today's AI for science remains a collaborator whose imagination and judgment benefit from human grounding.

cs.CY↗