arXiv subjects

Xiao Chen

Publications and source records attributed to Xiao Chen.

At least 37 records · Page 2Linked to original sources

arXiv2026

RAU: Reference-based Anatomical Understanding with Vision Language Models

Anatomical understanding, which is the ability to identify, localize, or segment anatomical structures, is critical in medical image analysis; however, its progress is constrained by the scarcity of expert-labeled data. A promising remedy is to leverage an annotated reference image to guide the interpretation of an unlabeled target. Although recent vision-language models (VLMs) exhibit non-trivial visual reasoning, their reference-based understanding and fine-grained localization remain limited. We introduce RAU, a framework for reference-based anatomical understanding with VLMs. We first show that a VLM learns to identify anatomical regions through relative spatial reasoning between reference and target images, trained on a moderately sized dataset. We validate this capability through visual question answering (VQA) and bounding box prediction. Next, we demonstrate that the VLM-derived spatial cues can be seamlessly integrated with the fine-grained segmentation capability of SAM2, enabling localization and pixel-level segmentation of small anatomical regions, such as vessel segments. Across two in-distribution and two out-of-distribution datasets, RAU consistently outperforms a SAM2 fine-tuning baseline using the same memory setup, yielding more accurate segmentations and more reliable localization. More importantly, its generalization ability to unseen modalities makes it scalable to unseen datasets, a property crucial for medical image applications. To the best of our knowledge, RAU is the first to explore the capability of VLMs for reference-based identification, localization, and segmentation of anatomical structures in medical images. Its promising performance highlights the potential of VLM-driven approaches for anatomical understanding in automated clinical workflows.

cs.CV↗

arXiv2026

Unifying Physics-Based Humanoid Interaction with a Context-Conditioned Interaction Prior

Developing unified physics-based humanoid controllers that can navigate complex 3D scenes and manipulate objects remains a longstanding challenge. Existing approaches are often specialized for either locomotion or object-centric manipulation, or rely on task-specific reward engineering that does not scale well across diverse behaviors. We present CHIP, a unified, physics-grounded framework for learning reusable humanoid interaction skills from heterogeneous motion data. Central to our approach is a conditional interaction prior that models a context-dependent distribution over these skills within a shared discrete space. Our method is trained in three stages. We first learn physics-based motion-imitation policies that acquire grounded teacher behaviors from heterogeneous interaction data. We then distill these behaviors into a context-conditioned interaction prior that captures reusable motion structure across locomotion and manipulation. Finally, we initialize downstream task policies from the pretrained prior and adapt them through prior-regularized online RL post-training. Experiments on a diverse suite of humanoid interaction tasks show that our approach supports scene-aware locomotion, contact-rich object manipulation, and compositional behaviors such as environment-aware object transport and long-horizon skill sequencing, while producing smooth transitions and physically plausible motion.

cs.RO↗

arXiv2026

Skill-Conditioned Gated Self-Distillation for LLM Reasoning

On-policy self-distillation (SD) improves LLM reasoning by using teacher-side privileged information (PI) to turn sparse verifier outcomes into dense token-level supervision. Existing methods usually assume trusted PI, such as reference answers or successful traces. We ask whether PI can instead come from an experience-derived skill bank, where retrieved skills are compact and reusable but may also be irrelevant or misleading. We propose Skill-Conditioned Gated Self-Distillation (SGSD), which formulates skill-based SD as teacher hypothesis validation rather than unconditional imitation. SGSD retrieves skill-mistake pairs, constructs a multi-teacher pool, and lets all skill-conditioned teachers score the same plain-prompt student rollout. The verifier validates each teacher's polarity: supporting a success or suppressing a failure gives positive supervision, while the opposite stance is reversed. A robust gated objective then distills informative teacher-student disagreements while suppressing uncertain or extreme signals. Experiments on multiple mathematical reasoning benchmarks show that SGSD consistently improves over GRPO and remains competitive with answer-conditioned OPSD under a weaker PI assumption. For example, on Qwen3-1.7B, SGSD outperforms GRPO by 6.2% and OPSD by 1.7% on average on AIME24, AIME25, and HMMT25.

cs.CL↗

arXiv2026

FUSE: An Evaluating Framework for Dangerous Capabilities of LLMs

Fragmented safety evaluation undermines the governance of dangerous AI capabilities. We present a modular framework that evaluates each model through three orthogonal pipelines---Knowledge ($K$), Defense ($D$), and Harm ($H$)---under a unified protocol, aggregating results into a standardized dangerous-capability profile $ϕ$. Pluggable modules supply scenario seeds, knowledge banks, hazard queries, and judge rubrics, while the core evaluation engine remains unchanged across domains; the CB evaluation is complemented by a cyber pilot demonstrating protocol transfer. Instantiating the framework with a chemical-biological (CB) module, we evaluate 12 commercial LLMs from four families. Our first contribution is a horizontal comparison of dangerous capability across models and model families: the three dimensions expose sharply divergent profiles---models with comparable knowledge differ in refusal resilience, and strong defenders do not generate less harmful content when they do comply---while family-level patterns further separate Claude, DeepSeek, and GPT models. The second is a temporal analysis of capability evolution: tracking $K$, $D$, and $H$ against model release dates reveals that dangerous capability has not monotonically declined; newer models deepen knowledge while only partially improving defense, showing that scaling and alignment progress do not uniformly translate into safety. Reliability is established via cross-judge consistency (bootstrap $ρ> 0.79$, 4 of 5 judges) and pipeline orthogonality ($K$--$D$--$H$ inter-correlations $ρ\in [0.32, 0.52]$).

cs.AI↗

arXiv2026

SpeechEditBench: A Bilingual Multi-Attribute Benchmark for Instruction-Guided Speech Editing

Instruction-guided speech editing requires a model to modify specified speech attributes while preserving non-target characteristics. Despite rapid progress in Speech Large Language Models (Speech LLMs), systematic evaluation of this capability remains challenging, as existing benchmarks are fragmented across isolated editing tasks. To bridge this gap, we introduce SpeechEditBench, a bilingual multi-attribute benchmark for instruction-guided speech editing. SpeechEditBench encompasses seven atomic editing tasks, as well as compositional editing tasks that integrate multiple operations within a single instruction. We propose an anchor-based evaluation protocol that separately assesses the edit success of target attributes and the preservation of non-target linguistic content, leading to three metrics: target success, preservation success, and joint success. Using this benchmark, we evaluate mainstream Speech LLMs and specialized speech editing systems. The results reveal three key findings: (1) no single model performs well across all editing dimensions; (2) closed-source Speech LLMs generally outperform open-source models; (3) compositional editing poses a significant challenge, with even the most advanced models struggling to achieve high joint success. SpeechEditBench provides a rigorous diagnostic framework to identify bottlenecks in Speech LLMs, thereby facilitating the development of next-generation models with more precise instruction-guided editing capabilities. Data and code are available at https://github.com/daxintan-cuhk/SpeechEditBench.

eess.AS↗

arXiv2026

Diffusive Relaxation of Participation Entropy in U(1)-symmetric Dynamics

Participation entropy (PE) quantifies the spread of a many-body wavefunction across configuration space. While PE relaxes rapidly in generic chaotic systems, we show that $\mathrm{U}(1)$ conservation laws slow it down by imprinting with the slow hydrodynamic modes. Using a cluster expansion around equilibrium, we show that, after local density inhomogeneities decay, the leading PE deficit is dominated by squared connected density correlations. The long time relaxation is therefore controlled by diffusive correlation spreading, giving $ΔS(t)\sim t^{-1/2}$ in the hydrodynamic regime and crossing over to $\sim \exp[-O(t/L^2)]$ when $t\geq L^2$. We confirm this entropy correlation relation using exact computation and infinite system tensor network simulations in various quantum $\mathrm{U}(1)$ conserving circuits. Our results establish PE as a sensitive probe of hydrodynamic memory and suggest that slow relaxation is a generic consequence of conservation laws.

quant-ph↗

arXiv2026

Cross-spectral Analysis of the Type-C Quasi-periodic Oscillation Shoulder Component in GX 339-4

We revisit Rossi X-ray Timing Explorer (RXTE) observations of GX~339$-$4 during the rising phase of its 2006/2007 outburst and apply a joint power-density-spectrum (PDS)--cross-spectrum (CS) decomposition to the type-C quasi-periodic oscillation (QPO) region. Within this framework, the QPO region is described by a narrow QPO fundamental and a neighboring high-frequency shoulder, whose amplitudes and phase lags can be measured separately. The shoulder is first detected at MJD~54142.04, mainly through the imaginary part of the CS and a narrow local structure in the phase-lag spectrum, before becoming a resolved high-frequency shoulder in the PDS. It follows the QPO frequency evolution on the high-frequency side, with $R_ν=ν_{\rm sh}/ν_{\rm QPO}\simeq1.04$--$1.18$. The QPO lag remains small, typically below $\sim0.17$~rad, whereas the shoulder carries a larger hard lag of $\sim0.5$--$0.8$~rad. Energy-resolved fits show the same separation: the QPO lag is close to zero or only weakly positive across most of the energy band, while the shoulder lag is systematically larger and generally increases with photon energy. The two components have broadly similar rms--energy shapes, although their relative strengths evolve during the observed sequence. Although the shoulder remains broad, with $Q\sim2$--$4$, its lag and rms--energy behavior resemble those of the type-B QPO detected shortly after our observations. This similarity raises the interesting possibility that the shoulder is related to an earlier, broader stage of the variability later seen as the type-B QPO.

astro-ph.HE↗

arXiv2026

Clustered Attractor Manifolds and Dynamical Condensation in Self-Attention

Transformer layers generate state-dependent interaction networks: token representations determine the attention matrix, which in turn updates the representations. We study this feedback in a minimal normalized self-attention dynamics and identify the overlap gap as the central quantity governing its attractor structure in the thermodynamic limit. When tokens form internally aligned clusters and their similarity to members of the same cluster exceeds that to every other cluster by a nonvanishing amount, inter-cluster attention is exponentially suppressed as the dimension increases. This mechanism produces a high-dimensional manifold of clustered fixed points, ranging from a few macroscopic clusters to extensive microscopic fragmentation, and also controls their stability against perturbations. Starting from an unstructured Gaussian state, we find that clustered states nucleate from the diffuse background only above a finite threshold in attention sharpness, giving rise to a dynamical attention-condensation transition.

cond-mat.dis-nn↗

arXiv2026

Respect Your Zero-Shot Uncertainty: Conservative Calibration for Test-Time-Adapted Vision-Language Models

Test-time adaptation (TTA) can improve the recognition accuracy of vision-language models under distribution shift, but often degrades calibration, making predictive confidence unreliable for downstream decision-making. Many existing label-free calibration approaches are either coupled to prompt optimization or rely on logit-range statistics that provide only a coarse characterization of the predictive distribution. We show that TTA can increase confidence and reduce entropy even when the top-1 prediction and its correctness remain unchanged, a failure mode we term prediction-preserving sharpening. Across diverse TTA methods and benchmarks, larger entropy reductions relative to paired zero-shot predictions are associated with greater increases in Expected Calibration Error (ECE). On entropy-reduced samples, confidence gains also tend to exceed accuracy gains. Based on these findings, we propose Zero-Shot-Anchored Entropy Calibration (ZAEC), a label-free post-hoc method that uses zero-shot entropy as a sample-specific uncertainty reference. ZAEC selectively restores the zero-shot entropy of sharpened predictions through minimal temperature scaling while leaving all other predictions unchanged. It requires no labeled calibration data or learned parameters and preserves class rankings and classification accuracy. Across five TTA methods and 15 datasets, ZAEC achieves the lowest post-hoc macro-average ECE on ViT-B/16, with consistent gains on RN50.

cs.CV↗

arXiv2026

Physics-Informed Quantum Machine Learning with Hard Constraint Embedding for Nonlinear Differential Equations of the First Order

Quantum algorithms based on linear-system approaches for solving differential equations demand qubit and precision resources beyond near-term capabilities. To address these challenges, this work proposes a physics-informed quantum machine learning (PIQML) framework with hard constraint embedding, specifically designed for NISQ era. Within this framework, parameterized quantum circuits serve as machine learning models, where the input variable is encoded into a high-dimensional feature space via a Fourier feature map. Subsequently, to eliminate approximation errors in critical physical conditions, the solution is constructed through a rigorously designed function mapper that analytically enforces initial conditions as hard constraints. Crucially, we compute derivatives with respect to the input variable using the parameter-shift rule---a quantum native gradient evaluation technique that avoids classical discretization. Unlike generic loss functions that target abstract data patterns, our loss function focuses on the differential equation residual and reference data. This design ensures that the trained model not only approximates the data but also intrinsically satisfies the physical constraint expressed by the DE itself. Our method is validated on several differential equations, including highly oscillatory ones, demonstrating its capability to tackle challenging nonlinear dynamics. Results demonstrate that our quantum model successfully learns the solution, showing close agreement with a high-precision classical numerical benchmark.

quant-ph↗

arXiv2026

CARE: A Cascaded Framework for Efficient and Reliable Time Series Anomaly Detection

While deep learning models have achieved state-of-the-art performance in time series anomaly detection, their complex architectures incur substantial inference overhead. Existing methods typically apply a uniform inference strategy across all data points, which is inefficient given that anomalies are inherently scarce and the vast majority of temporal data consists of predictable normal patterns. To mitigate this bottleneck, we propose CARE, a model-agnostic cascaded inference framework that integrates a Lightweight Pre-filter Model (LPM) with an existing high-capacity Complex Detection Model (CDM). The LPM rapidly filters high-confidence normal samples using a Residual MLP AutoEncoder and a Normality-Conditioned Gating mechanism. Crucially, we introduce a Structure Attention module to explicitly capture channel-wise anomaly contributions, and optimize the gating network via a confidence-guided selective routing objective that learns reliable routing decisions to reduce unnecessary CDM invocations. Extensive experiments across eight real-world benchmarks demonstrate that CARE effectively isolates high-confidence normal samples. By routing only uncertain samples to the CDM, our framework achieves $2.7\times$ to $4.8\times$ inference speedup compared to the most accurate SOTA approaches, while still maintaining competitive detection quality.

cs.LG↗

arXiv2026

SharedAssembly: A Data Collection Approach via Shared Tele-Assembly

High-precision, tight-clearance assembly demonstrations are indispensable for training tactile-aware robotic foundation models, yet their acquisition is heavily bottlenecked by the high operational barriers of conventional teleoperation. To bridge this gap, we propose SharedAssembly, a novel shared-autonomy bilateral teleoperation framework that embeds assembly-specific intelligence across both leader and follower sides to facilitate scalable data collection. Rigorous real-world user studies on challenging sub-millimeter tasks show that SharedAssembly achieves an exceptional 97% assembly success rate while significantly boosting completion efficiency. Notably, these performance gains become even more pronounced as the assembly clearance shrinks. Furthermore, our framework effectively eliminates the expertise gap, enabling novice operators to match or even outperform expert operators using conventional systems. By minimizing the skill barrier, SharedAssembly provides an efficient, robust, and accessible solution for large-scale data harvesting in contact-rich manipulation.

cs.RO↗

arXiv2026

AutoMat: Enabling Automated Crystal Structure Reconstruction from Microscopy via Agentic Tool Use

Reconstructing atomistic crystal structures from a single noisy STEM projection is an ill-posed inverse problem: multiple lattices can explain similar contrast, and purely feed-forward models cannot verify physical validity. We present AutoMat, a failure-aware agentic controller that performs inference-time hypothesis search with closed-loop verification to convert Scanning Transmission Electron Microscopy (STEM) images into simulation-ready crystal structures and downstream properties. AutoMat composes perception and physics modules---pattern-adaptive denoising, physics-guided template retrieval as a state-dependent auxiliary branch, symmetry-constrained atomic reconstruction, and MLIP-based relaxation/validation---and triggers rollback-and-retry when verification fails. For systematic evaluation, we introduce STEM2Mat-Bench, a benchmark dataset containing 450+ annotated samples. Performance is assessed using lattice root-mean-square deviation (RMSD), formation energy mean absolute error (MAE), and structure matching accuracy. Results demonstrate that AutoMat outperforms existing approaches including SOTA models, specialized domain tools, and closed-source multimodal large models. This work establishes a direct pathway from microscopic characterization to atomic-scale modeling, addressing a fundamental challenge in materials science.

cs.CV↗

arXiv2026

Scaling Behavior Foundation Model for Humanoid Robots

Humanoid control requires natural whole-body coordination, precise real-time responses to control signals, and robust generalization across diverse environmental contexts, making it a cornerstone for generalist embodied agents. Behavior Foundation Models (BFMs) have recently emerged as a promising solution to address these challenges by leveraging large-scale behavioral data to achieve superior expressiveness, versatility and generalization. However, despite growing interest in scaling BFMs to further improve their capabilities, it remains unclear how key factors, including the learning paradigm, behavioral data and model architecture should be coordinated to enable effective scaling. In this work, we revisit the scaling recipe for BFMs and demonstrate that substantial performance gains can be achieved through the coordination of three core components: 1) the learning paradigm of motion tracking that reformulates diverse humanoid control problems as the reproduction of integrated whole-body behaviors in the global frame; 2) the strategic synergy between on-policy rollout quantity and reference motion diversity; and 3) the expressive and scalable model architecture termed Humanoid Transformer that facilitates the natural emergence of structured behavioral representations. Through extensive experiments in both simulation and real-world deployment, we demonstrate that our approach yields significant improvements in control fidelity and task generalization, reducing Mean Per-Keypoint Position Error (MPKPE) on the test set by over 10% in local mode and 82% in global mode compared with existing humanoid controllers. These results establish BFM as a principled and effective foundation for scalable and general-purpose humanoid control.

cs.RO↗

arXiv2026

Coherent error induced phase transition

We investigate the stability of logical information in quantum stabilizer codes subject to coherent unitary errors. Beginning with a logical state, we apply a random unitary error channel and subsequently measure stabilizer checks, resulting in a syndrome-dependent post-measurement state. By examining both this \emph{syndrome state} and the associated syndrome distribution, we identify a phase transition in the behavior of the logical information. In the Clifford/stabilizer setting, the change of logical stabilizer structure in a syndrome branch determines the state-level MAP Pauli-frame return probability for a chosen logical-basis input. We separately use quantum coherent information to diagnose recoverability of arbitrary encoded inputs at the channel level. Below a critical error threshold $p_c$, the syndrome branches remain compatible with the original logical sector, enabling a high state-return probability. Above $p_c$, the data are consistent with effective scrambling of the encoded subspace relative to the stabilizer checks, where the syndrome-resolved logical maps approach a global-Clifford-induced stabilizer instrument. This common post-threshold picture organizes both the toric-code and finite-rate random-stabilizer-code results within the same scrambling mechanism. The syndrome distribution supplies complementary local and global structural diagnostics whose scaling depends on the code family. We refer to this phenomenon as a \emph{coherent error induced phase transition}. To illustrate this transition, we study two classes of quantum error-correcting codes, the toric code and finite-rate random stabilizer codes, thereby shedding light on the design and performance limits of quantum error correction under coherent errors.

quant-ph↗

arXiv2026

Damping Analysis of Subsynchronous Oscillations at the Sending-End of Practical Grid-Forming MTDC Power System for Isolated Renewable Energy

Subsynchronous oscillations (SSOs) have occurred at the sending-end of Zhangbei grid-forming MTDC power system (SE-GFPS). This paper gives an alert for the wide application of grid-forming (GFM) control by presenting a detailed report on such SSOs and the practical GFM configuration. Different from traditional studies that consider only parts of GFM control loops for simplification, power synchronous control (PSC), d- and q-axis AC voltage control (AVC), and inner current control (ICC) loops are comprehensively involved in this study. A self- and coupling-damping method is proposed to quantify the impact of both the inherent dynamics of different GFM control loops and the external dynamic coupling between the GFM control and the remaining SE-GFPS on the SSOs. It determines whether the major causes of SSOs are attributed to inherent GFM dynamics or external dynamic couplings. Based on the damping sensitivity analysis, the major impact factors of the SSOs are identified. Self-damping can be improved more by q-axis than by d-axis AVC parameters, and negative coupling damping can be reduced by PSC parameters. Finally, SSO mitigation strategies are proposed, and an SE-GFPS mirroring the real-world Zhangbei project is established on the electromagnetic transient (EMT) platform in Simulink, validating the accuracy of our conclusions.

eess.SY↗

arXiv2026

Timing and spectral analysis of the 2025 outburst of 4U 1630$-$47 with \textit{NICER}

We analyzed \textit{NICER} observations of the 2025 outburst of the black hole X-ray binary 4U~1630$-$47 to investigate the spectral--timing properties of its transient low-frequency quasi-periodic oscillations (QPOs) and millihertz-scale quasi-regular modulation (QRM). During the rising phase of the outburst, the QPO centroid frequency increased from $\sim 0.24$ Hz to $\sim 3.43$ Hz. Wavelet-based state separation shows that the with-QPO intervals are associated with a higher inner disk temperature and a lower \texttt{diskbb} normalization than the without-QPO intervals, while the photon index ($Γ$) shows weaker changes within the uncertainties. Near the outburst peak, the source displayed a weak QRM at $\sim 0.07$ Hz with a fractional rms amplitude of $\sim 4.7\%$, lower than that of the heartbeat state observed in 2023. Phase-resolved Hilbert--Huang analysis shows that the inner disk temperature is positively correlated with the X-ray flux, the \texttt{diskbb} normalization is anticorrelated, and $Γ$ varies only weakly. Overall, the short-timescale spectral--timing variability is expressed most clearly through the disk-related parameters. The transient QPOs are therefore consistent with short-timescale disk-related variability during the rising phase, whereas the millihertz-scale QRM may represent a weaker heartbeat-like variability mode appearing near the outburst peak.

astro-ph.HE↗

arXiv2026

Measurement-and Feedback-Driven Non-Equilibrium Phase Transitions on a Quantum Processor

Mid-circuit measurements and feedback operations conditioned on the measurement outcomes are essential for implementing quantum error-correction on quantum hardware. When integrated in quantum many-body dynamics, they can give rise to novel non-equilibrium phase transitions both at the level of each individual quantum trajectory and the averaged quantum channel. Experimentally resolving both transitions on realistic devices has been challenging due to limitations on the fidelity and the significant latency for performing mid-circuit measurements and feedback operations in real time. Here, we develop a superconducting quantum processor that enables global mid-circuit measurement with an average quantum non-demolition (QND) fidelity of 98.7% and fast conditional feedback with a 200 ns real-time decision latency. Using this platform, we demonstrate the coexistence of an absorbing-state transition in the quantum channel and a measurement-induced entanglement transition at the level of individual quantum trajectories. For the absorbing-state transition, we experimentally extract a set of critical exponents at the transition point, which is in excellent agreement with the directed percolation universality class. Crucially, the two transitions occur at distinct values of the tuning parameter. Our results demonstrate that adaptive quantum circuits provide a powerful platform for exploring non-equilibrium quantum many-body dynamics.

quant-ph↗