Search arXiv⌕ Search

arXiv subjects

Zheng Zhang

Publications and source records attributed to Zheng Zhang.

At least 73 records · Page 4Linked to original sources
arXiv2026

HandEdit: A Unified Benchmark for Egocentric Human-to-Robot Dexterous Hand Image Editing

Robotic manipulation with dexterous hands is a cornerstone of Embodied AI, yet its progress is stifled by the high cost of collecting embodiment-aware teleoperation data. While abundant egocentric videos of human hands offer a scalable alternative, the profound discrepancies in appearance, articulation, and camera viewpoints between human and robotic data raise significant challenges for co-training. Though existing general image-editing models demonstrate strong capabilities, they lack necessary embodiment-specific priors to fully bridge this gap. In this work, we present HandEdit, a unified large-scale embodiment-aware image-editing dataset and benchmark specifically designed to transform human hands and arms into various dexterous robotic embodiments within egocentric frames. HandEdit comprises over 200M editing instances derived from five diverse source datasets, covering 26 distinct URDFs, including 13 hand-only and 13 hand-arm configurations. Alongside the dataset, we establish a unified benchmark protocol with two tracks: Hand-only and Hand-Arm, supporting URDF-conditioned evaluation. We conduct extensive evaluations of 11 representative image-editing baselines using a multi-dimensional metric suite, including generic similarity metrics, VLM-based judgment, and embodiment-aware metrics. HandEdit serves as a critical resource at the intersection of image editing and robotics: it advances embodiment-aware editing models while enabling scalable dexterous robotic learning from abundant human video data, paving the way for more generalizable Embodied AI.

cs.RO↗
arXiv2026

RankGuide: Tensor-Rank-Guided Routing and Steering for Efficient Reasoning

Large reasoning models (LRMs) enhance problem-solving capabilities by generating explicit multi-step chains of thought (CoT) reasoning; however, they incur substantial inference latency and computational overhead. To mitigate this issue, recent works have explored model collaboration paradigms, where small reasoning models (SRMs) generate intermediate reasoning steps to achieve a better accuracy--latency trade-off. Despite recent progress, effectively and efficiently detecting and mitigating SRM failures in collaborative systems remains a key challenge. To address this issue, we analyze SRM inference in both the generated text and hidden-state spaces, and identify three types of failure modes: \textit{overconfidence}, \textit{uncertainty}, and \textit{heavy revalidation}. Building on these insights, we propose \textbf{RankGuide}, a framework that improves the efficiency and effectiveness of SRM--LRM collaboration through tensor-rank-guided routing and steering. Specifically, RankGuide leverages a routing signal that incorporates tensor-rank signals derived from consecutive hidden states to detect when SRMs are likely to fail and selectively invoke LRMs. In addition, we introduce a tensor-rank-filtered steering vector extraction method to modulate the reasoning trajectory of SRMs, thereby improving their generation quality. By improving both routing and steering through tensor-rank signals, RankGuide enables SRM--LRM collaborative systems to achieve more efficient reasoning with fewer steps and improved accuracy. Experiments across three reasoning domains -- mathematics, code generation, and scientific QA -- demonstrate the efficacy of RankGuide in reducing latency by up to $1.75\times$ compared to LRM, while maintaining competitive accuracy relative to prior methods. The code is available at \href{https://github.com/TTTTTTris/RankGuide}{https://github.com/TTTTTTris/RankGuide}.

cs.AI↗
arXiv2026

When Classes Evolve: A Benchmark and Framework for Stage-Aware Class-Incremental Learning

Class-Incremental Learning (CIL) aims to sequentially learn new classes while mitigating catastrophic forgetting of previously learned knowledge. Conventional CIL approaches implicitly assume that classes are morphologically static, focusing primarily on preserving previously learned representations as new classes are introduced. In practice, however, instances of the same semantic class may undergo substantial morphological evolution, such as a larva turning into a butterfly. Consequently, a model must both discriminate between classes and adapt to evolving appearances within a single class. To systematically address this challenge, we formalize Stage-Aware CIL (Stage-CIL), a paradigm in which each class is learned progressively through distinct morphological stages. We further introduce Stage-Bench, a 10-domain, two-stage benchmark and protocol for evaluating both inter-class forgetting and stage-level degradation within classes. Finally, we propose STAGE, an evolution-aware reference baseline that disentangles semantic identity from evolution dynamics through a fixed-size memory pool, enabling stage-aware prediction of later morphological forms from earlier representations. Extensive experiments show that conventional CIL reductions and existing continual-learning baselines remain insufficient under Stage-CIL, while STAGE consistently outperforms strong competitors, demonstrating the promise of explicit evolution-aware modeling for this new setting.

cs.LG↗
arXiv2026

CaM-Wolf: Causal-Aware Multimodal Agents for Social Deduction Games

Social deduction games (SDGs) such as Werewolf have become challenging testbeds for AI agents. These games require complex social skills such as reasoning, deception, and collaboration. While recent advances in large language models (LLMs) have driven significant progress in SDG agents, current approaches are predominantly text-based, overlooking the multimodal nature that is fundamental to human social interaction. To bridge this gap, we introduce CaM-Wolf, the first SDG agent that integrates multimodal perception and generation. CaM-Wolf processes video inputs from other players, employs a causal-aware Reasoner trained via reinforcement learning to establish logical chains between observable behaviors and hidden roles, and presents itself through an animated avatar. Our experiments and user study show that CaM-Wolf achieves superior agent gameplay performance and enhances the quality of human-AI interaction. This work represents a significant advancement towards creating more human-like AI agents capable of participating in nuanced social dynamics. Our code is available at https://3dagentworld.github.io/avatar_wolf.

cs.AI↗
arXiv2026

Detectors Learn the Wrong Thing: Shortcut-Resistant Adversarial Training Against Physically Realizable Attacks

AI-enabled visual perception systems are increasingly deployed in intelligent transportation infrastructure and autonomous vehicle related applications. However, physically realizable adversarial appearances pose a significant reliability challenge for these safety-critical systems. Adversarial training is effective, but repeated co-occurrence between adversarial texture and positive person instances can cause detectors to treat the texture itself as evidence of object presence, forming a patch texture shortcut. The detector may then treat texture as evidence for the target, causing false detections on texture-only inputs and weakening cross attack generalisation. We propose InsCAT, an instance-level contrastive adversarial training framework that prevents detectors from using adversarial texture as an independent decision cue. SICA aligns adversarial person features with matched clean features and separates them from texture-only negatives, while ROPO and Guard maintain online attack pressure and coordinate training. We evaluate eight independently generated attack textures on rendered nuScenes, INRIAPerson, printed garments, and three detector families. InsCAT achieves an average attack AP of 82.3% on rendered nuScenes, exceeding the strongest baseline by 11.1 points.Relative to AT-Mix, texture FPR decreases from 46.9% to 7.3%. Physical tests yield an F1 score of 96.6% and an FPR of 1.8%. Consistent gains across separately trained detectors demonstrate applicability across architectures with direct inference. The findings show that robust physical detection depends on preserving target related evidence while preventing adversarial texture from becoming an independent decision cu

cs.CV↗
arXiv2026

MuonQ: Enhancing Low-Bit Muon Quantization via Directional Fidelity Optimization

The Muon optimizer has emerged as a compelling alternative to Adam for training large language models, achieving remarkable computational savings through gradient orthogonalization. However, Muon's optimizer state is more sensitive to quantization errors: because the orthogonalization discards the magnitudes of singular values and retains only directional information, even small quantization errors in singular vector directions are amplified in the update. In this work, we propose MuonQ, a low-bit Muon training framework built on the principle of directional fidelity optimization. First, we apply a pre-quantization normalization so that each step introduces quantization errors of the same magnitude, preventing the accumulated error from developing a preferred direction. Second, we introduce a structural decomposition that separately quantizes the dominant singular components via power iteration, ensuring that quantization errors perturb only singular value magnitudes rather than rotating singular vector directions. Third, we adopt $μ$-law companding quantization to allocate higher resolution to densely packed momentum values, shifting the quantization objective from outlier preservation to dense-region distinguishability. Together, these techniques enable stable 4-bit quantization of Muon's optimizer states. Pre-training experiments on GPT-style and LLaMA-style models demonstrate that MuonQ at 4-bit precision recovers most of full-precision Muon's training loss and downstream accuracy while reducing optimizer-state memory by up to 7.3$\times$, offering a favorable point on the accuracy-memory trade-off. Our code is available at https://github.com/YupengSu/MuonQ.

cs.LG↗
arXiv2026

Dynamics of Long-lived Carriers in Molybdenum Carbide Nanosheets

Molybdenum carbide (MoC) is a promising candidate for substituting expensive platinum-group metals in many applications owing to its low cost and excellent properties, therefore a comprehensive understanding of its carrier dynamics is demanded for advancing the synthesis strategy and implementation of MoC. In this work, the carrier relaxation in MoC nanosheets is investigated by combining the femtosecond transient reflection spectroscopy with the first-principles calculations. The measured processes of electron-electron, electron-phonon, and phonon-phonon scattering show significantly longer lifetimes compared to those of other transition metal carbides. Through the analysis of calculated phonon dispersion, the nanosecond carrier lifetime is explained by the restricted phonon decay pathways, which is induced by the large mass difference between Mo and C atoms. The observed slow carrier cooling rate in our study offers a straightforward approach for effectively utilizing hot carriers, which is expected to improve the performance of photothermal and photovoltaic devices.

cond-mat.mtrl-sci↗
arXiv2026

Large Language Models Explore by Latent Distilling

Generating diverse responses is crucial for test-time scaling of large language models (LLMs), yet standard stochastic sampling mostly yields surface-level lexical variation, limiting semantic exploration. In this paper, we propose Exploratory Sampling (ESamp), a decoding approach that explicitly encourages semantic diversity during generation. ESamp is motivated by the well-known observation that neural networks tend to make lower-error predictions on inputs similar to those encountered before, and incur higher prediction error on novel ones. Building on this property, we train a lightweight Distiller at test time to predict deep-layer hidden representations of the LLM from its shallow-layer representations to model the LLM's depth-wise representation transitions. During decoding, the Distiller continuously adapts to the mappings induced by the current generation context. ESamp uses the prediction error as a novelty signal to reweight candidate token extensions conditioned on the current prefix, thereby biasing decoding toward less-explored semantic patterns. ESamp is implemented with an asynchronous training--inference pipeline, with less than 5% worst case overhead (1.2% in the optimized release). Empirical results show that ESamp significantly boosts the Pass@k efficiency of reasoning models, showing superior or comparable performance to strong stochastic and heuristic baselines. Notably, ESamp achieves robust generalization across mathematics, science, and code generation benchmarks and breaks the trade-off between diversity and coherence in creative writing. Our code has released at: https://github.com/LinesHogan/tLLM.

cs.CL↗
arXiv2026

DM-KG: A Novel Method for Boosting Spatial Cognition of Vision-Language Models in Street View Imagery

As vision-language models (VLMs) are increasingly deployed in geospatial question answering and visual scene understanding, improving their spatial cognition capability on street view imagery for complex logical reasoning has emerged as a key research priority. However, existing VLMs frequently suffer from "spatial semantic hallucinations" when perceiving object locations, distances, and directions in real-world street view scenes. Furthermore, such errors are often recalcitrant to tracing and calibration, posing a critical bottleneck for their practical deployment in geospatial tasks. To address this pressing challenge, this study proposes DM-KG (Direction-Metric Knowledge Graph), a structurally grounded spatial representation framework for street view imagery. By explicitly extracting directional and metric relationships between entities from a single 2D image, this framework enhances the spatial reasoning accuracy of VLMs through a structured knowledge graph. Specifically, we integrate panoptic segmentation with metric depth estimation to robustly compute entity-level 3D spatial coordinates. Subsequently, we encode the clock azimuths and Euclidean distances of entity pairs into a JSON-formatted knowledge graph, which is injected into the VLM as an explicit geometric prior to guide spatial reasoning. Experimental results on public spatial question-answering (QA) benchmarks demonstrate that DM-KG reduces the mean absolute error (MAE) in distance estimation by 31.1% and the mean angular error in direction judgment by 65.8%, while simultaneously maintaining a high QA success rate. By establishing a complete, augmented reasoning pipeline, this research significantly improves the spatial cognitive capabilities of VLMs in street view scenarios, thereby providing a flexible, generalized, and interpretable framework for geographic visual question answering (GeoVQA) in open environments.

cs.CV↗
arXiv2026

FlowPET: Physics-Informed Symplectic Flow Matching for Low-Count PET Reconstruction

Low-count Positron Emission Tomography (PET) reconstruction is severely hindered by the dissipative nature of prevailing generative models, where the inherent phase-space contraction leads to the numerical extinction (``wash-out'') of weak but diagnostically critical lesion signals. To overcome this geometric limitation, we propose \textbf{FlowPET}, a physics-informed framework that reformulates reconstruction as volume-preserving transport in a symplectic phase space. By parameterizing the posterior dynamics via a Separable Hamiltonian System, our approach guarantees a divergence-free vector field by construction, theoretically immunizing weak signals against probability mass collapse. To steer this conservative flow, we introduce conjugate boundary conditions based on the Range-Null space decomposition of the PET operator; this strictly enforces data consistency in the range space while confining stochastic uncertainty injection to the unobserved null space. We train the model via symplectic flow matching and perform inference using a symplectic leapfrog integrator. Extensive experiments on BrainWeb, clinical pediatric, and UDPET datasets demonstrate that \textbf{FlowPET} not only surpasses state-of-the-art deterministic and stochastic baselines in SSIM and PSNR but, more crucially, exhibits superior recovery of low-contrast lesions. The results confirm that imposing Hamiltonian structural constraints offers a robust geometric safeguard for medical inverse problems in high-noise regimes.

cs.CV↗
arXiv2026

Kairos: A Regret-Aware Native World-Action Model Stack for Physical AI

We introduce \textbf{Kairos}, a regret-aware native world-action model stack for Physical AI. Kairos is motivated by the view that a physical world model should not aim to fully simulate all future pixels, but should learn and maintain the information most relevant to embodiment control: object state, spatial relations, contact conditions, task progress, action consequences, failure boundaries, and deployment uncertainty. Kairos establishes three model-side prerequisites toward this goal. First, it \textbf{learns} control-relevant information through a \textbf{Cross-Embodiment Data Curriculum}, which organizes open-world videos, human behavioral data, and robot interactions into an intervention-strength progression from passive physical observation to intentional behavior and embodied action grounding. Second, it \textbf{maintains} control-sufficient states through a unified \textbf{understanding, generation, and prediction architecture} equipped with \textbf{Hybrid Linear Temporal Attention}, where local, mid-range, and global temporal pathways support multi-timescale state maintenance under efficient inference. Third, it \textbf{deploys} these states through a \textbf{Deployment-Aware System Co-Design}, treating latency, memory footprint, and hardware compatibility as first-order constraints for future observation, action, and feedback loops. Experiments on embodied world-model benchmarks, world-action benchmarks, long-horizon generation, and inference-efficiency evaluation show that Kairos achieves superior performance while offering a favorable efficiency to capability trade-off.

cs.AI↗
arXiv2026

One-dimensional carbon nanostructures with periodic graphitic nitrogen substitution

Heteroatom substitution is a powerful route to tune the chemical and electronic properties of carbon nanomaterials. In particular, replacement of an sp2 hybridized carbon atom in the graphene lattice with a nitrogen atom (denoted as graphitic nitrogen) induces substantial changes in the electronic properties. These include changes in the band structure that can influence electronic transport, and magnetism. A key requirement for applications is both the periodic and precise incorporation of the heteroatoms in extended carbon lattices. Here, we report the on-surface synthesis and characterization of two one dimensional carbon nanostructures, a polymer and a graphene nanoribbon, consisting of periodically incorporated graphitic nitrogen atoms. The on-surface reactions toward formation of the nanostructures were monitored by scanning tunneling microscopy. The bond-resolved chemical structures of the reaction intermediates and products were investigated by atomic force microscopy, which enabled atomic-scale visualization of the graphitic nitrogen sites. The electronic properties of the nanostructures were studied by scanning tunneling spectroscopy and density functional theory calculations. Our analyses revealed the presence of localized nitrogen-centered electronic states. In the gas phase where the nanostructures are in a neutral charge state, these states undergo spin polarization leading to an open-shell ground state. Upon adsorption on Au(111), the nanostructures exhibit electron transfer to the surface, which resulted in a closed-shell ground state. Our results demonstrate a straightforward and generally applicable route to synthesize graphitic nitrogen-substituted carbon nanomaterials with potential applications in spintronics, catalysis and energy storage.

cond-mat.mes-hall↗
arXiv2026

Continuous Speculative Decoding for Autoregressive Image Generation

Continuous visual autoregressive (AR) models have demonstrated promising performance in image generation, but their inherently sequential nature results in slow inference speed. Speculative decoding, a successful acceleration technique for large language models (LLMs), has effectively accelerated discrete visual AR models. However, the absence of an analogous theory for continuous distributions precludes its use in accelerating continuous AR models. To fill this gap, this work presents continuous speculative decoding, and addresses challenges from: 1) low acceptance rate, caused by inconsistent output distribution modeled by target and draft models, and 2) modified distribution without analytic expression, caused by a complex integral. For challenge 1), we address low acceptance rates through an approximated criterion, a novel denoising trajectory alignment strategy based on reparameterization proximity, and token pre-filling. For challenge 2), we introduce acceptance-rejection sampling algorithm with an appropriate upper bound, thereby avoiding explicitly calculating the integral. Furthermore, our denoising trajectory alignment is also reused in acceptance-rejection sampling, effectively avoiding repetitive diffusion model inference. Extensive experiments on various models at 256x256 and 512x512 resolutions demonstrate that our approach achieves over 2x wall-time speedup while preserving the image generation quality. Codes is available at: https://github.com/MarkXCloud/CSpD

cs.CV↗
arXiv2026

Formal Security Analysis of Agent Protocol Composition

AI agent protocols define how agents use tools, delegate work, and coordinate across software systems, but their security requirements remain incomplete and inconsistently enforced across deployments. We present AgentThread, a source-linked framework for security assurance analysis of agent protocols, from specification text to running SDKs. AgentThread contributes a layered security scope, protocol-derived checks formalized as TLA+ invariants, and a two-phase checker that compiles protocol specifications into model-checkable models and replays executable counterexamples against real SDKs through protocol adapters. For each finding, AgentThread records the source text behind the check and separates violated protocol requirements from missing recommendations, hardening gaps, and unassigned cross-protocol responsibilities. Across five emerging agent protocols, AgentThread identifies 35 specification-level findings, supports them with 80 implementation tests against production SDKs and reference servers, and finds 30 additional failures that emerge only under protocol composition. We further show that only one protocol enforces a security-relevant control in practice and no protocol assigns enforcement for cross-protocol behavior. Insecurity in agent protocols is therefore not only a specification or implementation problem, but also a responsibility gap across protocols, SDKs, and deployments.

cs.CR↗
arXiv2026

Probing Anomalous Microwave Emission with the Square Kilometre Array

Anomalous microwave emission (AME) represents an excess of radiation in the 10-60 GHz range, distinct from synchrotron, free-free, or thermal dust emission. Although most commonly attributed to electric dipole radiation from rapidly rotating small dust grains (spinning dust), alternative mechanisms such as magnetic dipole emission (MDE) remain plausible. The detection of AME across diverse environments, from diffuse interstellar clouds to protoplanetary disks and external galaxies, suggests that multiple physical processes or carriers may contribute to its origin. Understanding AME is essential for both Galactic astrophysics and cosmology, as it constitutes a significant foreground for cosmic microwave background (CMB) studies, potentially biasing measurements. This chapter reviews current theoretical frameworks and observational evidence for AME, highlighting the key outstanding questions concerning its emission mechanisms, carriers, and polarization properties. We discuss how the Square Kilometre Array Observatory (SKAO), through its unprecedented sensitivity, angular resolution, and frequency coverage, will transform AME studies. SKA observations will enable detailed mapping of AME morphology, precise characterisation of its spectral energy distribution, and the identification of its carriers in Galactic and extragalactic environments. By combining SKA-mid data with higher-frequency observations from ALMA and other facilities such as SPHEREx, it will be possible to disentangle competing models and exploit AME as a diagnostic probe of interstellar grain physics and the small-scale structure of the interstellar medium.

astro-ph.GA↗
arXiv2026

Methodological Frontiers in 21-cm Intensity Mapping: the Treatment of Systematics and Foreground Contamination

The distribution of neutral hydrogen (HI) in the post-reionization universe traces the cosmic large-scale structure and therefore serves as a powerful cosmological probe. An efficient way to measure its distribution over wide sky areas and redshift ranges is through single-dish intensity mapping, which exploits the autocorrelation signal of each dish in a telescope array while scanning the same sky patch. Thanks to its broad frequency coverage and technical capabilities, SKA-Mid will enable measurements of the integrated 21 cm emission from HI up to redshift $z\sim3$, making single-dish intensity mapping a key observable for probing dark matter and dark energy. Isolating the faint 21 cm cosmological signal without introducing biases is, however, challenging. The 21 cm signal is several orders of magnitude weaker than the astrophysical foregrounds, and its analysis is further affected by instrumental systematics. Overcoming these difficulties requires detailed modelling together with continuous improvements and innovations in data-analysis techniques. Over the past decade, the international community has developed and tested new methods to address current observational challenges and prepare for forthcoming SKA-Mid observations. This chapter reviews recent advances in map-making and component-separation techniques, with particular emphasis on telescope-specific systematics such as beam response and correlated noise. We focus on results obtained in controlled simulation environments, providing a valuable framework for assessing the strengths and limitations of different approaches. Developing robust algorithms capable of accurately handling instrumental effects and sky-model uncertainties is a crucial step toward fully exploiting the cosmological potential of HI intensity-mapping surveys in the SKA Observatory era.

astro-ph.CO↗
arXiv2026

Higher-Order Debiased Estimators for General Treatment Models

It is now well known that estimators based on influence functions can be sub-optimal in terms of convergence rates in various settings. To address this issue, higher-order influence functions (HOIF) are developed, generalizing the classical semiparametric theory. However, most existing results in this regard focus on treatment effect parameters defined in explicit forms, such as average treatment effects (ATE). In applications, economists are often confronted with tasks of inferring more complex parameters, such as quantile treatment effects (QTE) or effects of complicated treatment regimes/policy. These more complex parameters can often only be implicitly defined as the solution to nonlinear estimating equations, which correspond to M/Z-estimation problems. Our current understanding of these problems is mainly limited to the classical semiparametric theory. Given the foundational role of HOIF for estimating explicit parameters such as ATE, a modest step toward enriching the statistical foundation of econometrics and causal inference is to develop the corresponding higher-order estimators for those more complex parameters. To this end, we consider parameters of a class of non-separable structural models in the econometrics literature and develop a class of higher-order estimators for the target parameters. Statistical properties of these higher-order estimators are derived using recent advances in U-processes theory. Our proposed higher-order estimators relax complexity-reducing assumptions, quantified by Holder smoothness, imposed on the nuisance parameters compared to existing alternative estimators for many important parameters in this class, including QTE and quantile dose-response functions, among others.

econ.EM↗
arXiv2026

FirmCure:Towards Autonomous and Adaptive Rehosting of Linux-Based Firmware

Full-system rehosting plays a critical role in the security analysis of Linux-based firmware. It matches commonly deployed firmware with sufficient background knowledge. However, for custom devices, existing approaches struggle to handle initialization and runtime obstacles in the rehosting process caused by specialized architectures and hardware-dependent configuration, which heavily rely on expert intervention. This ultimately creates fundamental bottlenecks and results in low rehosting efficiency. To address the above challenges, we propose FirmCure, the first LLM-driven full-system rehosting framework designed for autonomous and adaptive rehosting of Linux-based firmware. FirmCure develops an Adaptive Perception Inference mechanism to extract firmware structural dependencies via static analysis, followed by a Reflective Synthesis module for iterative configuration optimization, and finally an Autonomous Runtime Intervention module for real-time error remediation through runtime fault diagnosis and monitoring. We evaluated 21 IoT firmware images from 10 vendors across 5 architectures, while FirmCure achieved a 100% network port opening rate and 90.5% service interactivity, substantially outperforming state-of-the-art baselines. Our experiments confirm that FirmCure's intervention strategies generalize across heterogeneous firmware. The framework successfully reproduces known vulnerabilities and discovers new security flaws.

cs.CR↗