Search arXiv⌕ Search

arXiv subjects

Zhuo Chen

Publications and source records attributed to Zhuo Chen.

At least 19 recordsLinked to original sources
arXiv2026

Every Little Helps: Building Knowledge Graph Foundation Model with Fine-grained Transferable Multi-modal Tokens

Multi-modal knowledge graph reasoning (MMKGR) aims to predict the missing links by exploiting both graph structure information and multi-modal entity contents. Most existing works are designed for a transductive setting, which learns dataset-specific embeddings and struggles to generalize to new KGs. Recent knowledge graph foundation models (KGFMs) improve cross-KG transfer, but they mainly exploit structural patterns and ignore rich multi-modal signals. We address these gaps by proposing a token-based foundation model (TOFU) for MMKGR, which exhibits strong generalization across different MMKGs. TOFU discretizes structural, visual, and textual information into modality-specific tokens. TOFU then employs a hierarchical fusion architecture with mixture-of-message mechanisms, aiming to process these tokens and obtain transferable features for MMKGR. Experimental results on 17 transductive, inductive, and fully-inductive MMKGs show that TOFU generally outperforms strong KGFM and MMKGR baselines, delivering strong performance on unseen MMKGs.

cs.CL↗
arXiv2026

MedEvo: Medical Reasoning Evolution through Knowledge Navigation

Medical multimodal reasoning models are typically trained on teacher-generated rationales, discarding the valid reasoning in students' own failed trajectories, while general-domain correction methods offer little guidance for errors in image reading, anatomical context, or differential diagnosis. We propose MedEvo, which uses medical knowledge, namely a tag taxonomy and error-type diagnosis, to navigate both trajectory repair and data selection. For each failed rollout, a gold-blind auditor locates the first consequential error and provides a minimal, answer-free hint; the same student retains the valid prefix and regenerates the erroneous suffix. After each round, MedEvo identifies capability gaps across medical tags on a held-out feedback set and prioritizes related questions for new rollouts and repairs. With supervised fine-tuning (SFT) alone, MedEvo raises Qwen3.5-9B mean accuracy across seven benchmarks from 65.57% to 69.12% (47.40% to 54.50% on the multimodal track of MedXpertQA (MedXpertQA-MM)), surpassing teacher-distillation SFT (66.93%). It also raises Qwen3.5-35B and Qwen3.5-397B mean accuracy from 67.91% to 73.04% and from 71.83% to 74.95%, respectively. Our analyses show that failed trajectories contain reasoning worth preserving, that repair success does not imply learning value, and that independent feedback is key to sustained improvement.

cs.AI↗
arXiv2026

ECG-RSI: Recursive Self-Improvement through Targeted Electrocardiogram Waveform Evolution

Electrocardiogram (ECG) models need supervision that targets evolving weaknesses and remains faithful to waveform evidence. We introduce ECG-RSI, a recursive self-improvement framework combining verified waveform synthesis with feedback-guided intervention search. Signal remeasurement verifies waveform edits; construction errors guide generator repair, while learner errors motivate interventions compared on a separate selection set. Across seven benchmarks, three rounds raise Qwen3.5-35B-A3B's mean question-answering accuracy from 34.57% to 56.04% and mean classification F1 from 24.27 to 34.03. Across the retained trajectory, the preferred supervision shifts from direct reinforcement to confusion-focused contrasts. Continued feedback correction yields diminishing external gains. These results motivate separating error diagnosis from intervention selection and checking whether gains on the feedback set transfer to independent benchmarks.

cs.AI↗
arXiv2026

Spark-Audio 1.0 Technical Report

This paper introduces Spark-Audio 1.0, a speech foundation model trained on domestically produced computing infrastructure. The model accepts text and speech inputs and generates text outputs. It supports multilingual (99 languages) and multi-dialect (202 dialects) speech recognition, speech translation, audio perception and understanding, and both text- and speech-based dialogue, marking a shift from merely ``hearing clearly'' to genuinely ``understanding'' speech. Architecturally, Spark-Audio 1.0 couples a 0.65B-parameter dense audio encoder with a 30B-A3B Mixture-of-Experts (MoE) large language model (LLM). The audio encoder is pretrained with representation-enhanced BEST-RQ and then refined through multi-task supervised training. The audio--LLM is trained through staged pretraining, followed by post-training that combines supervised fine-tuning with on-policy distillation. In total, the model is trained on 13 million hours of audio data together with large-scale text data. We evaluate Spark-Audio 1.0 on mainstream audio and text benchmarks, as well as on in-house test sets drawn from real-world application scenarios. Despite using substantially less audio training data than Qwen3.5-Omni-Flash, Spark-Audio 1.0 achieves comparable performance on most audio and text tasks, outperforms it in multilingual speech recognition, spoken dialogue, and real-world application scenarios, and approaches the performance of larger closed-source models. Nevertheless, there remains room for improvement in instruction following and complex audio reasoning. Finally, the successful training of Spark-Audio 1.0 on domestic hardware, including Huawei Ascend 910B and Hygon BW1000 chips, demonstrates the feasibility of a domestic technical pathway for building speech foundation models.

eess.AS↗
arXiv2026

Thermodynamics of Viscously Heated Circumplanetary Disks

Circumplanetary disks (CPDs) play an important role in the formation of gas giants and their satellites, yet remain poorly understood owing to the lack of observables and comprehensive theoretical models. Incorporating $α$-viscosity, a free-fall prescription of mass infall and boundary layer accretion, we develop an axisymmetric viscous radiation hydrodynamic (RHD) model. We evolve this model for different physical parameters using the RHD code {\tt Guangqi}, which solves RHD equations under the flux-limited diffusion approximation. In eight out of nine simulations, we obtain quasi-steady-state solutions and verify self-consistent energy conversion and transport. Overall, disk thickness increases with accretion rate and decreases with planet mass, consistent with the expected trend in conventional $α$-disk models. On the other hand, we highlight that the boundary layer plays a significant role in setting angular momentum transport and thermodynamics of the inner CPD, yielding ``hot'' ($s_{\rm real}>11\,k_{\rm{B}}\,m_{\rm{H}}^{-1}$) accreted gas, and potentially leading to dust sublimation. We identify a supersonic accretion flow, amounting to $2\%$--$5\%$ of the total accretion rate, located at high elevations above the CPDs with low angular momentum in our simulations. If the planet is massive ($\gtrsim3\,M_{\rm J}$), this low angular momentum flow impacting the planetary surface could generate observable H$α$ emission when the shock accretion rate is $\gtrsim5.5\times10^{-8}\, M_{\rm J}\,\rm{yr}^{-1}$.

astro-ph.EP↗
arXiv2026

Adv-Camou: An Emissivity-based Physical Dynamic Adversarial Patch for Infrared Attack

With the widespread application of infrared object detection technologies in military domains, intelligent infrared adversarial camouflage capabilities have become critical. Existing physical attack methods predominantly employ static thermal suppression strategies, suffering from limited environmental adaptability, insufficient attack robustness, and inadequate visual concealment. This study proposes Adv-Camou, an emissivity-based physical dynamic adversarial patch that achieves paradigm-shifting advancements from static to dynamic attacks through algorithm-material co-innovation. First, a multi-level loss function based on feature confusion mechanisms is designed to deceive detectors by attenuating target saliency and enforcing background feature alignment. Second, adaptive physical transformations and differentiable binarization optimization are integrated to address physical robustness and realizability constraints. A 6*4 array of reversible electrochromic emissivity device is developed and tested across a range of background scenarios, enabling scene-adaptive dynamic camouflage through independent unit control. Experiments demonstrate that this method achieves 95.59% attack success rate (ASR), reducing average precision (AP) by 60.51%. Physical validations show that static adversarial patches attain 76.55% average ASR under multi-view, multi-range, and multi-spectral conditions, while dynamic patches enhance ASR up to 91.55% in complex environments (wilderness/urban) through adaptive switching. The seamless digital-to-physical transfer via tunable emissivity devices transcends the limitations of traditional static camouflage by establishing a dynamic infrared regulation capability. This advancement enhances the battlefield survivability of military assets and demonstrates transformative potential for adaptive camouflage in practical combat scenarios.

eess.IV↗
arXiv2026

From Earth to Mars and Beyond: AI-Driven Cosmic-Ray Validation and Prediction with CorsikaAgent

We calculate the sea-level cosmic-ray muon flux with CORSIKA and develop CorsikaAgent, an auditable large-language-model-based system for configuring, executing, and validating air-shower simulations, and for predicting particle fluxes in unexplored environments. The physics calculation uses proton primaries described by the solar-minimum spectrum of Papini et al., with UrQMD at low interaction energies and either Sibyll 2.3e or QGSJET-III at high energies. The predicted momentum spectra at approximately 0 and 75 degrees are compared with experimental measurements of the atmospheric muon flux and with commonly used analytic parametrizations. The two interaction-model configurations give similar spectra and describe the principal momentum and zenith-angle dependences of the data. The agent independently reconstructs the same observable through a different sampling strategy and keeps every steering card, executable, seed, log, and analysis script bound to the result it produced; its output is consistent with the direct calculation at the level resolved by the present comparison. Transferred to Mars, the workflow reproduces the CORSIKA 8 benchmark, and transferred to Venus it produces the first prediction of the surface muon flux, an environment for which no measurement exists. The agent thus extends the workflow from auditable reproduction to first-principles prediction: it autonomously reconfigures the simulation(atmosphere, magnetic field, electric field)for new environments, while the proton-only primary model, limited high-momentum statistics, and an incomplete systematic-uncertainty treatment bound the physical precision that can currently be claimed.

hep-ph↗
arXiv2026

TagBO: LLM-Driven Task-Aware Graph Bayesian Optimization for Scientific Discovery

Mixed-variable optimization in scientific discovery often requires exploring large combinatorial spaces under costly experimental evaluations, making sample?efficient Bayesian optimization (BO) essential. However, existing BO methods typ?ically rely on structural descriptors to define similarity among categorical variables, implicitly assuming that structural similarity reflects similarity in task performance. In scientific optimization, this assumption is often violated, leading to surrogate representations that are poorly aligned with the optimization objective. To address this limitation, we propose TagBO (Task-aware Graph Bayesian Optimization), a method that integrates LLM-derived relational priors with descriptor-based struc?tural information for mixed-variable Bayesian optimization. Rather than directly using LLM outputs as predictive scores, TagBO treats LLM outputs as a potentially noisy source of task-aware relational priors and extracts coarse ordinal relational signals from LLM reasoning to construct task-aware similarity structures for scien?tific optimization. TagBO then performs constrained Gaussian process Bayesian optimization over the resulting joint discrete-continuous representation space. Ex?periments on diverse chemistry and materials formulation tasks demonstrate that TagBO consistently improves sample efficiency under fixed evaluation budgets across LLMs of varying capabilities.

cs.LG↗
arXiv2026

TGuide: Faster Autoregressive Video Diffusion via Historical Trajectory Guidance

Can autoregressive video diffusion be accelerated further when only four denoising steps remain? Existing feature caching approaches struggle in this regime, as large feature changes between denoising timesteps undermine both direct reuse and lightweight predictors. We observe that features across video chunks remain strongly correlated at matching denoising timesteps, offering a complementary source of guidance. Building on this insight, we introduce TGuide, a framework that accelerates autoregressive video diffusion through a lightweight Anchor-Transport-Correct (ATC) predictor guided by historical sampling trajectories. Our ATC combines two complementary references: the current chunk's preceding-step features preserve its spatial structure, while the previous chunk's same-timestep features provide information at the target denoising stage. By integrating these references through spatial retrieval and adaptive correction, ATC improves velocity prediction accuracy despite the large gaps between denoising steps. To mitigate error accumulation across both denoising steps and video chunks, we train the predictor on its own autoregressive rollouts, allowing it to adapt to states shaped by its own earlier predictions. An optional confidence-aware scheduler further adjusts the quality-efficiency trade-off. Experiments on Self Forcing, Causal Forcing, and HY-WorldPlay show that TGuide preserves generation quality while achieving denoising speedups of up to $1.89\times$ over the original 4-step generators. On Self Forcing, TGuide generalizes from 81-frame training videos to 333-frame generation, approximately four times the training length, while exceeding the original model's VBench score.

cs.CV↗
arXiv2026

DiffMorphGrasp: A Diffusion Framework Conditioned on Hand Morphology for Cross-Embodiment Grasp Generation

As robots are deployed across diverse environments, manipulation tasks require robot hands with different morphologies to satisfy varying task requirements, hardware capabilities, and cost constraints. However, transferring grasping policies learned on one robot hand to another with a different morphology remains a fundamental challenge due to variations in kinematic structures and degrees of freedom across embodiments. To address these limitations, we propose DiffMorphGrasp, the first diffusion-based framework that incorporates hand morphology information into the generative process for cross-embodiment dexterous grasping. The proposed approach maps grasps from diverse robotic hands into a reference hand pose, providing a common space for learning. Grasp generation is then continuously conditioned on structured representations of hand kinematics, encoded as graphs derived from hand configurations, together with object geometry. In addition, we introduce a loss function that computes joint-level errors over the hand kinematic structure to constrain the generated grasps during training. To evaluate DiffMorphGrasp, we construct a new benchmark for assessing grasp generation performance across robot hands with different morphologies. Extensive experiments demonstrate that DiffMorphGrasp achieves strong zero-shot generalization to unseen hand morphologies, enabling scalable cross-embodiment grasp deployment. Project page: https://georgewuzy.github.io/DiffMorphGrasp-website

cs.RO↗
arXiv2026

Unravelling the Nature of AT 2025abao: From Precursor to Final Fate

Luminous red novae (LRNe) are potential signatures of common envelope evolution (CEE), a critical yet poorly understood phase in compact binary formation. AT 2025abao, an LRN in the Andromeda galaxy, originated from an asymptotic giant branch (AGB) progenitor and exhibited a prolonged pre-outburst transient. Lacking direct constraints on the companion, however, the system's evolutionary fate remains unclear. Here, we unravel the binary nature of AT 2025abao by modeling its complete evolution, from precursor to final outcome. We show that the eight-year infrared precursor arises from a $\approx1.4^{+0.2}_{-0.3}\,M_{\odot}$ main-sequence companion inspiralling through the wind of a $\approx 6.9\,M_{\odot}$ AGB star. As CEE commences, the companion plunges into the envelope and launches a shock. Radiation hydrodynamic modeling of this outburst constrains the total ejecta mass to $\approx 0.92\,M_{\odot}$ and energy to $\approx 1.25\times 10^{47}\,$erg, revealing that only $\approx 0.57\,M_{\odot}$ becomes unbound. Subsequent radiative transfer simulations successfully reproduce the evolving spectral energy distribution. Crucially, our models indicate the main-sequence companion and AGB core survive in an eccentric orbit without merging at least 400 days after CEE onset. AT 2025abao may therefore be the first observed CEE event caught forming a close compact binary, bridging a long-standing observational gap in compact binary formation.

astro-ph.SR↗
arXiv2026

VolCo: Volumetric Contact for High-Fidelity Human Grasp Generation

Accurate contact modeling is fundamental to understanding hand-object interaction, yet existing contact representations are typically restricted to object surfaces and rely on hand-crafted rules to recover contact details, leading to severe penetrations and implausible results. To better exploit the rich detail in motion-capture data, we introduce Volumetric Contact (VolCo), a representation that expands surface points to a set of 3D volumetric grids. VolCo encodes 3D contact that allows precise hand part recovery, and is organized in an inherent hierarchy: local contact details within each volume and global hand geometry across all volumes. Our framework, VolCoDiff, employs two modules to capture local and global features following this hierarchy. For local contact details, we use a 3D variational autoencoder to model the possible hand configurations conditioned on the local object signed distance field (SDF). For global hand geometry, we design a prior-guided diffusion model that learns the distribution of compressed latent features aggregated from the volumetric grids. We evaluate our method on two benchmark datasets and demonstrate state-of-the-art performance in penetration and stability, indicating the capability to generate tight grasps with much less severe penetrations. Our code is available at https://github.com/chzh9311/volco.

cs.CV↗
arXiv2026

RobotWorld: Benchmarking Multimodal Agents for Robot Use Across Diverse Tasks and Embodiments

General-purpose agents increasingly write code, use tools, and complete complex digital tasks, raising the question of how far these capabilities carry into the physical world. To investigate this, we introduce RobotWorld, a challenging simulation testbed for robot use: turning instructions and observations into physical task execution through robot interfaces. Its 84 tasks span manipulation, mobile manipulation, locomotion, driving, and aerial control, with explicit interaction budgets and executable success checks. By analysing task outcomes alongside execution traces, we identify both the capabilities that transfer and the gaps that prevent reliable completion. Furthermore, we find that current agents can construct sophisticated perception and control workflows, including image segmentation, camera calibration, spatial estimation, and dynamics-based computation. These capabilities, however, do not consistently compose into successful behaviour: agents lose task-relevant object states despite reaching commanded poses, fail to correct ineffective actions, recover too late, or mistake unfinished tasks for completion. This uneven transfer also differs across models: Astra succeeds more often on spatial and constrained-contact goals, whereas Opus 5.5 succeeds more often on continuous-balance and timed-interaction goals. By linking these outcomes to execution behaviour, RobotWorld provides both a rigorous proving ground and an empirical account of the remaining capability gaps, thereby establishing concrete targets for training and designing more reliable physical-world agents.

cs.RO↗
arXiv2026

Resonance expansions and local-energy decay estimates for Dirac operators

We study resonance expansions and localised long-time dynamics for three-dimensional semiclassical Dirac operators. For smooth Hermitian matrix-valued perturbations which are analytic outside a compact set, we first obtain exponential cut-off resolvent bounds away from the resonance set and polynomial bounds in thin resonance-free rectangles adjacent to the real axis. Combining these estimates with a local $\Oc(h^{-3})$ upper bound on the number of resonances, a pigeonhole selection of admissible contours, almost-analytic functional calculus, and a Cauchy--Green deformation, we derive resonance expansions for spectrally localised propagators. Positive-energy resonances contribute for large positive times, whereas negative-energy resonances contribute for large negative times. We then specialise to mass-type perturbations $\mbfβV(x)$ in a nontrapping energy window. Using the logarithmic resonance-free region established in our earlier work together with a quantitative FBI-transform/escape-function argument, we obtain a polynomial bound for the continued cut-off resolvent and deduce rapid local-energy decay. We subsequently refine this polynomial estimate by a finite-time propagation argument in the undistorted region. Introducing the maximal positive-sheet connection time $T_χ$ associated with the spatial cutoff, we prove a geometric continued-resolvent estimate and, for every fixed $M$ below the resonance-free depth constant, derive the explicit local-energy decay threshold $T_N>T_χ+(N+1)/M$. The same analysis gives a rigorous microlocal obstruction to a uniform $[h\log(1/h)]^{-1}$ bound for the full distorted resolvent and identifies a finite-flight-time amplification mechanism for the sandwiched continued resolvent at logarithmic depth.

math.AP↗
arXiv2026

Lower bound on Dirac resonances with mass-type potentials in the semi-classical limit

We establish lower bounds on the number of resonances for three-dimensional semiclassical Dirac operators with mass-type potentials. For a pair $\Dl_j=\Dl_0+\mbfβv_j$, $j=1,2$, satisfying the exterior analyticity, decay and non-degeneracy assumptions of the paper, we associate a relative distribution $μ$ to the corresponding mass functions. We prove that if $E_0>1$ is a boundary point of $\suppμ$, then every complex neighbourhood of $E_0$, and likewise of $-E_0$, contains in total at least $C\hbar^{-3}$ resonances of the two operators, counted with multiplicity. The proof transfers an analytic wave-front singularity of $μ$ to the relative classical density of states and combines this with Khochman's local trace formula and a relative semiclassical trace formula. As a consequence, if $v$ is a single bounded mass-type potential, analytic with power decay in an exterior sector, with $\inf(1+v)>0$ and $\sup v>0$, then $\Dl=\Dl_0+\mbfβv$ has at least $C\hbar^{-3}$ resonances in every complex neighbourhood of $E_0=1+\sup v$ and of $-E_0$, for all sufficiently small $\hbar$. The exponent agrees with Khochman's upper bound, the two resonance families are related by spectral symmetry, and $E_0$ is necessarily a trapping energy. We also prove the relative semiclassical trace formula needed in the argument for general Hermitian matrix-valued perturbations.

math.AP↗
arXiv2026

Breaking the Illusion of Review Reliability under Static Evaluation: SCOPE Fuzzing for LLM-based Scientific Reviewers

The rapid growth of submissions and reviewing workload has accelerated the use of large language models (LLMs) in peer review. Prior studies suggest that LLM-based reviewers can penalize content perturbations, such as overclaiming, indicating a certain degree of reliability. Yet these conclusions are largely based on a narrow set of perturbation strategies instantiated with static templates, providing limited evidence of actual reliability. In this paper, we construct a three-level evaluation framework covering perturbations to surface presentation, argumentative logic, and value judgment. Experiments on representative LLM-based reviewers reveal two limitations of static evaluation: stratified vulnerability, where perturbation effects depend on whether the paper's original review score is high or low, and perturbation undercoverage, where a single template misses vulnerabilities exposed by diverse realizations. To address these limitations, we propose SCOPE-Fuzzer, a strategy-aware fuzzer that combines feedback-driven strategy selection with adaptive mutation of paper content. By iteratively probing reviewers with dynamic perturbations, SCOPE-Fuzzer consistently uncovers vulnerabilities overlooked by static evaluation and other baselines.

cs.AI↗
arXiv2026

Information Greenhouse: Optimal Persuasion for Medical Test-Avoiders

Patients often avoid medical tests because the information they provide, although medically useful, is psychologically painful. This paper studies optimal communication between a doctor and an information-avoidant patient who can refuse testing and treatment. I characterize when optimal communication creates an information greenhouse, a commitment to reward participation with comforting information about the untreated prognosis. When testing is voluntary and the patient is unwilling to be tested under extreme pessimism, an information greenhouse is optimal and takes the form of committed comfort, which provides reassuring information after the test. When the patient can reject the consultation at the outset and the patient's prior belief about the untreated prognosis is intermediate, an information greenhouse is optimal and takes the form of precautionary comfort, which provides reassuring information before the test. In all other cases in which the patient can be persuaded, warning-based policies that trigger pessimism prevail.

econ.TH↗
arXiv2026

Accelerator and Brake: Dynamic Persuasion with Dead Ends

This paper studies dynamic persuasion in a strategic-experimentation relationship in which the principal has a single-peaked preference over the agent's stopping time. Excessive experimentation may end in a dead end. The principal privately observes project quality, which determines the agent's payoff conditional on success, while both parties learn about feasibility only through the agent's experimentation. We show that an optimal policy uses at most two one-shot disclosures: an accelerator before the principal's ideal stopping time and a brake afterward. A local Arrow--Pratt comparison of induced payoffs over stopping time determines whether the accelerator is concentrated or gradual. Under common discounting, the comparison yields a one-shot accelerator. Under heterogeneous discounting, the one-shot result remains robust unless the agent is sufficiently more impatient than the principal, in which case the ranking reverses over an interval and the accelerator can take a one-shot--gradual--one-shot form.

econ.TH↗