arXiv subjects

Xiao Zhang

Publications and source records attributed to Xiao Zhang.

At least 19 recordsLinked to original sources

arXiv2026

Asynchronous Replanning in Two Population Linear Quadratic Mean Field Games: Information Requirements and Stability

We study asynchronous replanning in a two population linear quadratic mean field game in which the populations may begin from different beliefs and hence use different plans. Each population observes its own aggregate trajectory and a public record of implemented revisions, while its continuation best response depends on the opponent's current plan. We identify the information required for replanning as the aggregate state at the end of the initial observation interval together with the opponent's active continuation plan. For linear observations, recoverability of this state-plan pair is characterized by a kernel inclusion, and a bounded factorization quantifies sensitivity to observation error. In particular, the required pair may be recoverable even when the full hidden belief is not. Once initialized, the public event record and the common best-response map recursively determine subsequent opponent plans, and the resulting local algorithm reproduces an ideal benchmark on every finite opportunity prefix; implemented revisions alternate as a consequence of best-response persistence. For finite populations, we derive an eventwise linear recursion for sampling errors, obtain finite prefix error bounds, and prove record matching for an autonomous deadband rule under a positive decision margin. Finally, we separate unique solvability of mutual continuation responses from stability of alternating responses, and show that at a pre-terminal Zeno accumulation, spectral stability together with a moving-boundary estimate yields convergence of the continuation plans to the equilibrium restarted from the actual limiting state.

math.OC↗

arXiv2026

Wearable Multimodal Human-Machine Interface for Integrated Hand Intentions Decoding in Dynamic Teleoperation

Under ubiquitous teleoperation environments with optically challenging conditions, an interface for tele-operated grasping that combines wearability with precise decoding of hand intentions (hand pose, gestures, and grasping force) is essential. Yet, existing interfaces often fall short in meeting these demands, compromising either the diversity of multiple intentions decoding or wearability. To address this, we developed a novel Multiple Intentions Decoding Human-Machine Interface (MI-DHMI) that integrates high-throughput surface electromyography (sEMG) sensors with hand-mounted and forearm-mounted inertial measurement units (IMUs). The developed interface is supported by a unified framework for simultaneous multiple intentions decoding. By employing multimodal deep learning and hardware design with a low noise floor, the decoding framework selectively focuses on the sEMG components that are genuinely associated with finger movements. This effectively reduces decoding errors caused by sEMG variability during unconstrained upper-limb motions, thereby significantly enhancing robustness. Even under unconstrained wrist and forearm motion, the interface achieves a gesture recognition accuracy exceeding 97%, grasping force estimation with $R^2 = 0.95$, and hand pose decoding consistent with the actual hand pose, outperforming baseline devices and algorithms. Ablation studies further validate the effectiveness of the proposed decoding framework. Finally, two online experiments were conducted to validate the device, demonstrating its superior performance in high-stability tasks, including a pouring task and object grasping. The developed interface provides a new solution of a fully wearable, multiple intentions decoding system, offering effective support for ubiquitous teleoperation and contributing to the advancement of human-machine interaction research.

cs.RO↗

arXiv2026

Thinking with Cameras: Active Visual Reasoning via Dynamic Viewpoint Control for Surveillance Video Understanding

Large vision-language models (LVLMs) have recently achieved remarkable progress in general-purpose video understanding. However, their application to surveillance videos remains challenging due to the lack of large-scale domain-specific datasets and the limitation of passive observation from fixed viewpoints. In surveillance scenarios, critical visual evidence can be easily missed when targets are distant, small, occluded, or move beyond the current camera view. In this work, we introduce CamVLM, a new framework for Thinking with Cameras, which enables LVLMs to actively acquire visual evidence through dynamic viewpoint control rather than passively analyzing fixed video streams. We first construct CCTV-Anomaly, a large-scale surveillance video understanding dataset containing 14,459 videos across 10 anomaly categories, with detailed captions and event annotations. We further formulate viewpoint control as an active visual perception problem and build CamTrack-53K, an object-centric viewpoint trajectory dataset for learning camera actions. Moreover, we propose a reinforcement learning based viewpoint policy optimization framework, which models camera control as a sequential decision-making process and learns long-horizon observation strategies beyond supervised trajectory imitation. Extensive experiments demonstrate that CamVLM achieves state-of-the-art performance under both passive observation and dynamic viewpoint settings, validating the effectiveness of active camera-based reasoning for surveillance video understanding. Our datasets, model, and code will be available at https://github.com/xiaozhang79/CamVLM .

cs.CV↗

arXiv2026

Empowering VLMs for Few-Shot Multimodal Time Series Classification via Tailored Agentic Reasoning

In this paper, we propose the first VL\underline{\textbf{M}} \underline{\textbf{a}}gentic \underline{\textbf{r}}easoning framework for few-\underline{\textbf{s}}hot multimodal \underline{\textbf{T}}ime \underline{\textbf{S}}eries \underline{\textbf{C}}lassification (\textsc{MarsTSC}), which introduces a self-evolving knowledge bank as a dynamic context iteratively refined via reflective agentic reasoning. The framework comprises three collaborative roles: i) Generator conducts reliable classification via reasoning; ii) Reflector diagnoses the root causes of reasoning errors to yield discriminative insights targeting the temporal features overlooked by Generator; iii) Modifier applies verified updates to the knowledge bank to prevent context collapse. We further introduce a test-time update strategy to enable cautious, continuous knowledge bank refinement to mitigate few-shot bias and distribution shift. Extensive experiments across 12 mainstream time series benchmark datasets demonstrate that \method{} delivers substantial and consistent performance gains across 5 VLM backbones, outperforming both classical and foundation model-based time series baselines under few-shot conditions, while producing interpretable rationales that ground each classification decision in human-readable feature evidence. Code is available at https://github.com/HuangJW0821/MarsTSC.

cs.AI↗

arXiv2026

Task-Level Natural Language Priors as Learning Signals for Low-Resource LLM Training

Large language models (LLMs) often struggle when low-resource training data are ambiguous or incomplete. Task-level natural-language priors can provide useful guidance in such settings, but existing approaches usually treat these priors as input context rather than as learning signals during training. We propose Prior-Guided Tuning (PGT), a training perspective that incorporates natural-language priors as auxiliary learning signals for low-resource LLM training. Under this perspective, we introduce Contrastive Prior Steering (CPS), which keeps the original supervised objective intact while adding positive and negative prior-conditioned auxiliary losses to encourage task-consistent learning and discourage plausible but misleading alternatives. Experiments on AmbiMath, Jigsaw, and MNLI/HANS show that CPS consistently improves over plain and prompt fine-tuning. On AmbiMath, CPS achieves 97.6% average exact-match accuracy. On Jigsaw, CPS improves average Macro F1 by 9.5 percentage points over standard fine-tuning, and with 1/10 of the experimental training data slightly exceeds full-data plain fine-tuning. On HANS, CPS improves non-entailment accuracy by 8.3 and 5.2 percentage points for LLaMA 3.1 8B and Qwen 2.5 7B, respectively, while maintaining comparable in-domain MNLI accuracy. These results support our central claim: task-level natural-language priors can provide useful guidance as auxiliary learning signals for low-resource LLM training. Our code and data will be publicly available.

cs.AI↗

arXiv2026

Partially Observed Mean Field Games Without Perfect Recall: Optimality Conditions and Equilibria

This paper studies partially observed mean field games without perfect recall (WPR). The representative agent observes a noisy signal, but the control at time $t$ uses only $\mathcal G_t^I=σ(y_t)$, a generally non-nested information family. The conditional population law instead uses the observation filtration $\mathbb F^Y$. These coupled levels rely on different information scales and are difficult to close within one construction. We parameterize the environment by a deterministic compatible joint law of state, driving variables, and random mean field term, thereby preserving its dependence structure without enlarging the agent's control information. For a fixed law, a reference measure and Girsanov's theorem yield a WPR stochastic maximum principle; the selected response is represented by the conditional Hamiltonian and WPR belief measure. The joint path posterior of hidden state and mean field term gives a weak Kushner-Stratonovich representation of the conditional population law. A recursive response map is continuous on a compact convex set of compatible laws, so Schauder-Tychonoff yields a weak WPR equilibrium. For a fixed equilibrium law and feedback, a compatible Yamada-Watanabe theorem lifts pathwise uniqueness to a strong realization. Finally, a linear-quadratic interbank lending example compares perfect recall (PR) with WPR. The PR response follows the Kalman-Bucy feedback, whereas the WPR response solves a Fredholm-Volterra equation and is affine in the current observation under Gaussianity. The numerical experiment illustrates how equilibrium behavior differs between PR and WPR.

math.OC↗

arXiv2026

Data-Centric Neuromotor Interfaces for Portable Human-Machine Interaction

Dexterous human-machine interaction requires intuitive and expressive interfaces that can be efficiently deployed on constrained edge devices. Flexible material-based neuromotor interfaces hold considerable promise, as they decode human movement intention into natural control. Although emerging flexible electronic skins enable wearable high-fidelity data acquisition, practical deployment inevitably involves trade-offs between computational resources and portability. We present a data-centric paradigm where physiological features yield fundamental separability, providing sufficient discriminative cues for recognition. A wireless, high-bandwidth system developed for collecting various electrophysiological signals, when integrated with muscle-specific electrodes, forms a surface electromyography-based interface. Exploiting highly separable data, a 2,210-parameter model achieves 94.36% accuracy across 34 gestures and can be rapidly deployed on edge devices, establishing a new thousand-parameter benchmark for dexterous decoding. The underlying data-algorithm interactions in the data-centric paradigm are further clarified, demonstrating its feasibility in real-world scenarios. This study provides a principled and validated pathway for practical deployment of reliable neuromotor interfaces.

cs.RO↗

arXiv2026

Peeling property for the Einstein scalar field equations with the nonzero cosmological constant

Inspired by interaction of gravitational waves, dark matters and dark energy, we study the Einstein scalar field equations with the nonzero cosmological constant for the Bondi-Sachs metrics. We provide the asymptotic expansions under the outgoing radiation condition. Unlike the case of the zero cosmological constant, the asymptotic expansions depend on four additional $Λ$-independent functions $B(u, θ, ϕ)$, $\tilde{X}(u, θ, ϕ)$, $\tilde{Y}(u, θ, ϕ)$ and $\tilde{I}(u)$. We show that, under suitable coordinate transformation, we can make $B(u, θ, ϕ)=0$. We prove the peeling property for the Einstein scalar field equations if $\tilde{I}=0$, and derive a loss formula of the Bondi energy-momentum for the nonzero cosmological constant. We remark that, for certain real data from gravitational waves, the loss formula is likely to indicate the loss property of the Bondi energy-momentum.

gr-qc↗

arXiv2026

GREAT: Generalizable Backdoor Attacks in RLHF via Emotion-Aware Trigger Synthesis

Recent work has shown that RLHF is highly susceptible to backdoor attacks. However, existing methods often rely on rare tokens or fixed triggers, limiting their impact in realistic scenarios. In this work, we develop GREAT, a novel framework for crafting natural distributional backdoors in RLHF. Specifically, GREAT targets harmful response generation for a vulnerable user subpopulation featured by semantically violent requests paired with emotionally angry triggers. At the core of our framework is a trigger identification pipeline that operates in the model's latent embedding space, leveraging dimensionality reduction and clustering techniques to identify representative triggers. To enable this, we introduce a hierarchical and diversity-driven prompting strategy to construct Erinyes, a high-quality dataset of over 5,000 angry triggers curated from GPT-4.1. Our experiments show that GREAT significantly outperforms baselines in attack generalization to unseen triggers, while preserving standard utility and maintaining stealth under defenses.

cs.CR↗

arXiv2026

Linear Quadratic Mean Field Games under Heterogeneous Erroneous Initial Information

We study a finite-horizon linear--quadratic mean field game with heterogeneous observations of the initial mean field. These observations may be erroneous and are used by agents to construct their feedback laws. At the population level, the heterogeneous error profile is generally infinite-dimensional. We derive exact linear sensitivity formulas showing that its closed-loop effects nevertheless admit a finite-dimensional closure: for each agent, the deviations are governed by only two n-dimensional error channels, its private error E_i and the population-average error \bar E. The population-average error determines the displacement of the actual mean field, whereas centered private errors determine agent-specific deviations and the additional cross-sectional covariance. The same representation yields quadratic cost identities and a uniform $O(N^{-1})$ mean-square approximation of the empirical aggregate. In the deterministic model, the private state history induces a linear observation problem for the initial errors. Nonsingularity of the associated observability Gramian is necessary and sufficient for exact recovery. The recovered errors reconstruct the actual mean-field state at the revision time and initialize a revised correct-information continuation from that state. We obtain an exact continuation-cost comparison; when the population-average error vanishes, revision does not increase the continuation cost. For stochastic dynamics, we take a two-component revision signal as given and study the resulting one-shot feedback relative to an oracle continuation initialized with the actual current mean field. The actual mean-field deviation is driven only by the population-average signal error, while individual deviations also retain the private signal error. Exact covariance and quadratic continuation-cost identities quantify these effects.

math.OC↗

arXiv2026

LongWoF-Bench: Evaluating EvoMap Genes for Verifiable Long-Workflow Tasks

Large language models are increasingly expected to execute complex workflows whose success depends on maintaining interdependent constraints and producing artifacts that satisfy strict end-to-end verification. Yet successful execution experience is typically lost after a single run, forcing subsequent models to rediscover strategies and failure modes from scratch. We study whether such experience can instead be externalized and reused through EvoMap, where verifier-confirmed execution trajectories are consolidated into structured Gene. To evaluate this setting, we introduce the Long-Workflow Benchmark (LongWoF-Bench), comprising 778 machine-verifiable tasks across code generation, agent-environment synthesis, mathematical reasoning, and rule following. On the 252 tasks with verifier-confirmed Opus trajectories, evolved EvoMap Gene outperform Skill across all seven evaluated models by 8.7-15.5 percentage points, with the gains extending to consumer models from different model families. In contrast, reference-distilled Gene do not exhibit the same advantage, indicating that compact representation alone is insufficient and that Gene utility is closely associated with verified experience provenance. For Claude Opus, Gene reuse also completes 39 more tasks than Skill while reducing solve-time token consumption by 9.9%. Together, these results show that verified execution experience can be retained and shared as a reusable external resource, enabling models to improve long-workflow completion without repeatedly paying the full cost of experience discovery.

cs.CL↗

arXiv2026

Prime multipliers of order ten: norm spectra, local charts, and selective lifting

Write P(N) for the assertion that every principal minor of the Fourier matrix of order N is nonzero. The square-free principal-minor conjecture was previously known for a few uniform small-multiplier families, including several families with two prime factors; broader higher-factor results were non-uniform, apart from isolated exact verifications. We prove a near-complete prime-multiplier theorem for the composite base 10: P(10p) holds for every prime p \notin {2,5,11}. The proof begins with the complete cyclotomic norm spectrum of the principal minors of the order-ten Fourier matrix. Its rational-prime support is {2,3,5,11,31}. Ordinary finite-characteristic lifting settles the norm-safe multipliers. A known small-multiplier theorem handles one norm-exceptional case; another is settled by retaining the prime-ideal chart in which each carrier degenerates. This yields a flag-selective lifting lemma and an active-rank norm budget. The chart analysis also isolates the limitation at the remaining square-free exception: active carriers can cover every local chart, so the first-order argument stops. We discuss this obstruction, the non-uniformity of fixed-base lifting, and the difficulties in passing to general square-free orders. All finite calculations are exact and have been independently confirmed.

math.NT↗

arXiv2026

Shadows and photon spheres of static black holes embedded in a Dehnen-(1,4,5/2)-type dark matter halo with a quintessential field

This paper investigates the appearance characteristics of static black holes embedded in Dehnen-(1,4,5/2)-type dark matter halos with a quintessential field, focusing on how the dark matter halo and dark energy affect the black hole images. We first derive the event horizon radius and the photon effective potential of the black hole, and then calculate critical quantities such as the critical photon sphere radius and critical impact parameter under different parameter sets. Trajectories of photons are subsequently plotted. The study reveals that as the parameters of the dark matter halo (the central density of the dark matter halo $ρ_s$ and the scale radius of the central halo $r_s$) and the quintessential field (the normalization factor $c$ and the equation of state parameter of dark energy $w_q$) increase, the aforementioned physical quantities generally exhibit an increasing trend. Based on the derived general expressions for the redshift factor and integrated intensity, we further explore the optical effects of the spherical accretion and the thin-disk accretion models. The results indicate that dark energy exerts an influence on the black hole shadow that is strongly dependent on the observer's position, whereas the influence exerted by dark matter exhibits no such conspicuous dependence. Furthermore, dark matter and dark energy have distinct effects on both the intensity and the radius of the black hole shadow. In particular, the intensity exhibits a greater sensitivity to dark energy, whereas the radius is more responsive to dark matter. This distinction offers a potential observational criterion for identifying, through black hole images, whether the dominant interacting component near the black hole is dark matter or dark energy, and provides an important basis for constraining the equation-of-state parameter $w_q$.

gr-qc↗

arXiv2026

Multimodal Adaptive Control for Safe Robotic Craniotomy Under Partial Observability

Autonomous robotic craniotomy requires continuous regulation of tool-tissue interactions to mitigate mechanical overload and thermal damage while maintaining surgical efficiency. However, this process is inherently partially observable due to unknown, time-varying tissue properties and the inability to directly measure cutting temperatures under physical occlusion. To address these challenges, we propose RL-MACRO, a cybernetic closed-loop intelligence framework that couples multimodal perception, adaptive decision-making, and robotic execution. This framework empowers the surgical robot to autonomously perceive inaccessible states from partial sensory feedback and dynamically optimize its behaviors under uncertain environment. A CNN-LSTM observer first fuses force and sound feedback to reconstruct the hidden temperature state (R^2=0.939, MAE = 1.717 deg C). This reconstructed temperature, alongside multi-sensor features, forms the belief state for an offline Implicit Q-Learning (IQL) policy. A novel dual-head Actor dynamically coordinates the feed rate, spindle speed, and cutting depth to optimize efficiency within strict safety bounds. These decisions are seamlessly translated into spatial motions via online trajectory re-planning and velocity servoing. Experiments on bovine ribs and six ex vivo goat skulls validate the system's robust perception, adaptive recovery from force/temperature excursions, and smooth execution on irregular surfaces, establishing a data-driven cybernetic paradigm for safe and efficient autonomous bone cutting.

cs.RO↗

arXiv2026

AutoResearch: Insight In, Hallucination Out

Autonomous research systems are increasingly capable of executing long research workflows, yet automation alone does not ensure that the resulting process remains scientifically grounded. We introduce AutoResearch, a two-stage system that connects Idea Generation with Idea Execution to address both how research ideas are formed and how they are reliably established through experimentation. In Idea Generation, AutoResearch continuously integrates emerging research signals with accumulated domain knowledge, identifies transferable mechanistic insights, and uses multi-model generation and cross-review to produce grounded, testable research plans. In Idea Execution, coordinated agents decompose these plans into experiments, iteratively implement and diagnose them, and employ independent evidence-based review before accepting research conclusions. Across representative settings in cross-modal retrieval, systems optimization, and benchmark-driven machine learning, AutoResearch turns generated ideas into measurable progress, detects and corrects unreliable experimental results, and makes evidence-conditioned decisions to continue, revise, or terminate research directions. For example, on RSICD benchmark, an AutoResearch-generated idea improves mean Recall from 32.84 to 34.69, while recording only 5 audit-confirmed issue events compared with 11-27 for other autonomous research systems. These results demonstrate a research process in which meaningful insight is grounded before experimentation and conclusions are grounded before acceptance: Insight In, Hallucination Out.

cs.AI↗

arXiv2026

HarnessSafe: Evaluating Safety Across Persistent Carriers in Agent Harnesses

Modern agent harnesses persist state across tasks and sessions through persistent carriers like memory, skills, tools, and shared artifacts. However, this capability creates delayed safety risks: attacker-influenced content can cross system boundaries and later affect the execution of a benign request. Existing benchmarks typically focus on a few carriers or harnesses, while end-to-end attack-success rates reveal little about how risks propagate. To this end, we present HarnessSafe, a benchmark comprising 328 executable cases across seven persistent-carrier families and evaluated on most mainstream agent harnesses. Each case is specified as a Persistent-Risk Lifecycle that traces attacker influence from its initial entry, through persistence across carriers and system boundaries, to a later benign trigger and an observable violation. We further introduce a multi-stage, trace-based evaluation that uses observable execution evidence to determine how far each attack chain progresses and where it is stopped. Experiments show that containment is carrier-specific and strongly depends on the harness-model configuration. Both the harness and model backend substantially shape containment outcomes, while attack success rates cannot reflect distinct lifecycle progression patterns.

cs.CR↗

arXiv2026

ChainClaw: A Layered Agent Framework for Reliable On-Chain Execution

General-purpose large language model agents have achieved strong performance on tool-augmented tasks, yet they rely on assumptions break down in blockchain environments. On-chain execution is stateful, adversarial, and economically irreversible, exposing three fundamental gaps: Reactivity, Irreversibility, and Observability. We propose ChainClaw, a blockchain-native agent framework built on OpenClaw, that addresses all three gaps through a layered architecture comprising an event-driven orchestration layer, a simulation-based safety intelligence layer, and an on-chain monitoring runtime layer, unified by a cross-layer memory subsystem. ChainClaw closes the Reactivity gap via event ingestion and simulation feedback, the Irreversibility gap via a pre-execution safety pipeline with transaction simulation and action guard, and the Observability gap via an on-chain read adapter and transaction monitor. We evaluate ChainClaw on a purpose-built benchmark covering seven tasks across four categories and five dimensions. ChainClaw consistently outperforms representative baselines on both safety and task completion.

cs.AI↗

arXiv2026

Optical properties of Ag, Au, and Cu from first principles

We present a comprehensive framework for investigating the optical response of metals from first principles that combines density functional theory, many-body perturbation theory, and efficient interpolation techniques based on maximally localized Wannier functions, and apply it to analyze the optical properties of silver (Ag), gold (Au), and copper (Cu). We evaluate the optical properties of these metallic materials considering both single-particle direct and phonon-assisted excitations, as well as the resistive Drude contribution. We find an overall excellent agreement with experimental optical measurements for these materials, and show that both single-particle and collective excitations are important in capturing their optical response in the infrared. Our methodology provides fundamental understanding of the optical response of metals and is generally applicable to investigate the optoelectronic properties of emerging metallic materials.

cond-mat.mtrl-sci↗