Searcharxiv⌕ Search

arXiv subjects

Chao Wang

Publications and source records attributed to Chao Wang.

At least 109 records · Page 6Linked to original sources
arXiv2026

Can LLM-Generated Text Empower Surgical Vision-Language Pre-training?

Recent advancements in self-supervised learning have led to powerful surgical vision encoders capable of spatiotemporal understanding. However, extending these visual foundations to multi-modal reasoning tasks is severely bottlenecked by the prohibitive cost of expert textual annotations. To overcome this scalability limitation, we introduce \textbf{LIME}, a large-scale multi-modal dataset derived from open-access surgical videos using human-free, Large Language Model (LLM)-generated narratives. While LIME offers immense scalability, unverified generated texts may contain errors, including hallucinations, that could potentially lead to catastrophically degraded pre-trained medical priors in standard contrastive pipelines. To mitigate this, we propose \textbf{SurgLIME}, a parameter-efficient Vision-Language Pre-training (VLP) framework designed to learn reliable cross-modal alignments using noisy narratives. SurgLIME preserves foundational medical priors using a LoRA-adapted dual-encoder architecture and introduces an automated confidence estimation mechanism that dynamically down-weights uncertain text during contrastive alignment. Evaluations on the AutoLaparo and Cholec80 benchmarks show that SurgLIME achieves competitive zero-shot cross-modal alignment while preserving the robust linear probing performance of the visual foundation model. Dataset, code, and models are publicly available at https://github.com/visurg-ai/SurgLIME.

cs.CV↗
arXiv2026

Fast Text-to-Audio Generation with One-Step Sampling via Energy-Scoring and Auxiliary Contextual Representation Distillation

Autoregressive (AR) models with diffusion heads have recently achieved strong text-to-audio performance, yet their iterative decoding and multi-step sampling process introduce high-latency issues. To address this bottleneck, we propose a one-step sampling framework that combines an energy-distance training objective with representation-level distillation. An energy-scoring head maps Gaussian noise directly to audio latents in one step, eliminating the need for a costly recursive diffusion sampling process, while distillation from a masked autoregressive (MAR) text-to-audio model preserves the strong conditioning learned during diffusion training. On the AudioCaps benchmark, our method consistently outperforms prior one-step baselines such as ConsistencyTTA, SoundCTM, AudioLCM and AudioTurbo, on both objective and subjective metrics, while substantially narrowing the quality gap to AR diffusion systems with multi-step sampling. Compared to the state-of-the-art AR diffusion system, IMPACT, our approach achieves up to $8.5$x faster batch inference with highly competitive audio quality. These results demonstrate that combining energy-distance training with representation-level distillation provides an effective recipe for fast, high-quality text-to-audio synthesis.

cs.SD↗
arXiv2026

Efficient Symbolic Execution of Software under Fault Attacks

We propose a symbolic execution method for analyzing the safety of software under fault attacks both accurately and efficiently. Fault attacks leverage physically injected hardware faults in an embedded system to break the safety of a software program. While there are existing methods for analyzing the impact of maliciously injected hardware faults on the embedded software, they suffer from inaccurate fault modeling and inefficient fault analysis. To overcome these limitations, we propose two novel techniques. First, we propose a new fault modeling technique that leverages automated program transformation to add symbolic variables to the original program, to accurately model the new program behavior induced by the injected faults. This new fault modeling approach has two advantages over existing techniques: (a) the fault-induced program behavior is closely related to what attackers exploit in practice and (b) the automatically transformed program may be analyzed by any downstream fault analysis algorithm. Second, we propose an efficient symbolic execution algorithm that is designed specifically for conducting fault analysis on the transformed program. It leverages two pruning techniques to mitigate path explosion. We have implemented the proposed method and evaluated it on a variety of benchmark programs. The experimental results show that our method significantly outperforms the state-of-the-art techniques. Compared to the current state-of-the-art, it is able to detect previously-missed safety violations and at the same time avoid bogus violations. Furthermore, compared to the baseline algorithm, our optimized symbolic execution algorithm can be orders-of-magnitude faster.

cs.SE↗
arXiv2026

Distributional Off-Policy Evaluation with Deep Quantile Process Regression

This paper investigates the off-policy evaluation (OPE) problem from a distributional perspective. Rather than focusing solely on the expectation of the total return, as in most existing OPE methods, we aim to estimate the entire return distribution. To this end, we introduce a quantile-based approach for OPE using deep quantile process regression, presenting a novel algorithm called Deep Quantile Process regression-based Off-Policy Evaluation (DQPOPE). We provide new theoretical insights into the deep quantile process regression technique, extending existing approaches that estimate discrete quantiles to estimate a continuous quantile function. A key contribution of our work is the rigorous sample complexity analysis for distributional OPE with deep neural networks, bridging theoretical analysis with practical algorithmic implementations. We show that DQPOPE achieves statistical advantages by estimating the full return distribution using the same sample size required to estimate a single policy value using conventional methods. Empirical studies further show that DQPOPE provides significantly more precise and robust policy value estimates than standard methods, thereby enhancing the practical applicability and effectiveness of distributional reinforcement learning approaches.

stat.ML↗
arXiv2026

Tree Training: Accelerating Agentic LLMs Training via Shared Prefix Reuse

Agentic large language model (LLM) training often involves multi-turn interaction trajectories that branch into multiple execution paths due to concurrent tool use, think-mode, sub-agent, context management and other runtime designs. As a result, the tokens produced by a single task naturally form a tree-structured token trajectory with shared prefixes, rather than a linear sequence. Existing training pipelines linearize such trajectories and treat each branch independently, leading to substantial redundant computation in both forward and backward passes. We derive that averaging the loss over all branches independently is algebraically identical to a per-token weighted loss, where each token's weight equals the fraction of branches passing through it. The problem therefore reduces to computing the log-probability of every token in the prefix tree exactly once, with no repeated computation across shared prefixes: we propose DFS serialization of the tree, which visits every token exactly once, and adapt full-attention and SSM layers to ensure the resulting log-probabilities match independent per-branch calculation exactly. In practice, a single trajectory tree can be too large to fit in GPU memory; we therefore propose Redundancy-Free Tree Partitioning, which handles memory-constrained settings with zero redundant computation and peak memory bounded by a single root-to-leaf path. Together, these contributions form Tree Training, an efficient framework for training LLMs on tree-structured trajectories, achieving up to 6.2x end-to-end training speedup on dense and MoE models for both supervised fine-tuning and reinforcement learning.

cs.LG↗
arXiv2026

Relative Entropy Estimation in Function Space: Theory and Applications to Trajectory Inference

Trajectory Inference (TI) seeks to recover latent dynamical processes from snapshot data, where only independent samples from time-indexed marginals are observed. In applications such as single-cell genomics, destructive measurements make path-space laws non-identifiable from finitely many marginals, leaving held-out marginal prediction as the dominant but limited evaluation protocol. We introduce a general framework for estimating the Kullback-Leibler divergence (KL) divergence between probability measures on function space, yielding a tractable, data-driven estimator that is scalable to realistic snapshot datasets. We validate the accuracy of our estimator on a benchmark suite, where the estimated functional KL closely matches the analytic KL. Applying this framework to synthetic and real scRNA-seq datasets, we show that current evaluation metrics often give inconsistent assessments, whereas path-space KL enables a coherent comparison of trajectory inference methods and exposes discrepancies in inferred dynamics, especially in regions with sparse or missing data. These results support functional KL as a principled criterion for evaluating trajectory inference under partial observability.

cs.LG↗
arXiv2026

LKV: End-to-End Learning of Head-wise Budgets and Token Selection for LLM KV Cache Eviction

Long-context inference in Large Language Models (LLMs) is bottlenecked by the linear growth of Key-Value (KV) cache memory. Existing KV cache compression paradigms are fundamentally limited by heuristics: heuristic budgeting relies on statistical priors rather than task objectives, causing resource misallocation, while heuristic selection relies on coupled query-key interactions or static inductive biases (e.g., attention sinks). To address this limitation, we introduce LKV (Learned KV Eviction), which formulates KV compression as an end-to-end differentiable optimization problem. LKV integrates LKV-H to learn task-optimized global budgets, and LKV-T to derive intrinsic KV importance without materializing attention matrices. This design bypasses heuristic proxies, strictly aligning compression with task objectives. Extensive evaluations demonstrate that LKV achieves state-of-the-art performance on both LongBench and RULER benchmarks at high compression rates. In particular, on LongBench, LKV achieves near-lossless performance with only 15\% KV cache retention. Crucially, our analysis identifies learned budgeting as the dominant driver of fidelity, demonstrating that data-driven allocation is essential to overcome the limitations of hand-crafted heuristics.

cs.LG↗
arXiv2026

Learning Evolution via Optimization Knowledge Adaptation

The iterative search process of evolutionary algorithms (EAs) encapsulates optimization knowledge within historical populations and fitness evaluations. Effective utilization of this knowledge is crucial for facilitating knowledge transfer and online adaptation. However, current research typically addresses these goals in isolation and faces distinct limitations: evolutionary sequential transfer optimization often suffers from incomplete utilization of prior knowledge, while adaptive strategies, utilizing real-time knowledge, are limited to tailoring specific evolutionary operators. To simultaneously achieve these two capabilities, we introduce the Optimization Knowledge Adaptation Evolutionary Model (OKAEM), a unified learnable evolutionary framework capable of adaptively updating parameters based on available optimization knowledge. By parameterizing evolutionary operators via attention mechanisms, OKAEM enables learnable update rules that facilitate the utilization of optimization knowledge via two phases: pre-training to integrate extensive prior knowledge for efficient transfer, and adaptive optimization to dynamically update parameters based on real-time knowledge. Experimental results confirm that OKAEM significantly outperforms state-of-the-art sequential transfer methods across 12 transfer scenarios via pre-training, and surpasses advanced learnable EAs solely through its self-tuning mechanism in prior-free settings. Beyond demonstrating practical utility in prompt tuning for vision-language models, ablation studies validate the necessity of the learnable components, while visualization analyses reveal the model's capacity to autonomously discover interpretable evolutionary principles. The code can be accessed at https://gitee.com/Anonymity_Paper/code-of-okaem.

cs.NE↗
arXiv2026

Nonlinear stability of 2-D Couette flow for the compressible Navier-Stokes equations at high Reynolds number

In this paper, we investigate the nonlinear stability of the Couette flow for the two-dimensional compressible Navier--Stokes equations at high Reynolds numbers ($Re$) regime. It was proved that if the initial data $(ρ_{in},u_{in})$ satisfies $\|(ρ_{in},u_{in})-(1, y, 0)\|_{H^4(\mathbb{T}\times\mathbb{R})}\leq εRe^{-1}$ for some small $ε$ independent of $Re$, then the corresponding solution exists globally and remains close to the Couette flow for all time. Formal asymptotics indicate that this stability threshold is sharp within the class of Sobolev perturbations. The proof relies on the Fourier-multiplier method and exploits three essential ingredients: (i) the introduction of ``good unknowns" that decouple the perturbation system; (ii) the construction of a carefully designed Fourier multiplier that simultaneously captures the enhanced dissipation and inviscid-damping effects while taming the lift-up mechanism; and (iii) the design of distinct energy functionals for the incompressible and compressible modes.

math.AP↗
arXiv2026

A Unified Poisson Summation Framework for Generalized Quantum Matrix Transformations

We present a unified algorithmic framework for quantum simulation of non-unitary dynamics and matrix functions, governed by the principle of spectral aliasing derived from the Poisson Summation Formula (PSF). By reinterpreting discretization errors as spectral folding in dual domains, we synthesize two distinct algorithmic paths: (i) the Fourier-PSF path, generalizing transmutation methods for time-domain filtering, which is optimal for singular and fractional dynamics $e^{-tH^α}$, here $H\succeq 0$; and (ii) the contour-PSF path, a novel discrete contour transform based on the resolvent formalism, which achieves exponential convergence for holomorphic matrix functions via radius optimization. This dual framework resolves the smoothness-sparsity trade-off: it utilizes the Fourier basis to handle branch-point singularities where analyticity fails, and the Resolvent basis to exploit complex-plane regularity where it exists. We demonstrate the versatility of this framework by efficiently simulating diverse phenomena, from fractional anomalous diffusion to high-precision solutions of stiff differential equations, outperforming existing methods in their respective optimal regimes.

quant-ph↗
arXiv2026

Too Private to Tell: Practical Token Theft Attacks on Apple Intelligence

Apple Intelligence is a generative AI (GenAI) service provided by Apple on its devices. While offering a similar set of features as other similar GenAI services, Apple Intelligence is claimed to be designed with an extra focus on user security and privacy through a two-stage authentication and authorization design using anonymous access tokens. In this paper, we present our investigation into this token issuance mechanism with a goal to reveal possible vulnerabilities using traffic analysis, reverse engineering, and cross comparison with Apple's public documentation. Specifically, we present the Serpent attack, the first practical cross-device token replay attack against Apple Intelligence that allows the attacker to steal the access tokens from the victim's device and utilise them on a different device, with all usage rate-limited against the victim. We have achieved successful attacks on the latest macOS 26 Tahoe and demonstrated that an attacker, who even has used up its own allowance, can immediately regain access to Apple Intelligence service. We have responsibly disclosed the vulnerabilities to the vendors and received confirmation from Apple with CVE assigned and bounty given. Our results highlight a general lesson for built-in AI services: Anonymising identity does not by itself make the AI service secure; Enforcing non-transferability requires cryptographic binding to the rightful user.

cs.CR↗
arXiv2026

Calibrated Speculative Decoding: Frequency-Guided Candidate Selection for Efficient Inference

Speculative decoding accelerates autoregressive generation by letting draft tokens bypass full verification, but conventional frameworks suffer from frequent false rejections, particularly when draft models produce semantically correct but lexically divergent outputs. In this paper, we present Calibrated Speculative Decoding (CSD), a training-free framework that recovers valid tokens discarded by standard verification. Guided by the principle of "Frequency-Guided Candidate Selection and Probability-Guarded Acceptance," CSD incorporates two lightweight modules: Online Correction Memory, which aggregates historical rejections to propose recurring divergence patterns as rescue candidates, and Semantic Consistency Gating, which verifies candidate admissibility using probability ratios instead of exact token matching. Our evaluation across diverse large language models demonstrates that CSD outperforms existing methods, achieving a peak throughput speedup of 2.33x. CSD preserves model accuracy across all tasks while further boosting performance on complex reasoning datasets. These results establish CSD as a highly effective, lightweight solution for practical LLM deployments.

cs.CL↗
arXiv2026

ClipGStream: Clip-Stream Gaussian Splatting for Any Length and Any Motion Multi-View Dynamic Scene Reconstruction

Dynamic 3D scene reconstruction is essential for immersive media such as VR, MR, and XR, yet remains challenging for long multi-view sequences with large-scale motion. Existing dynamic Gaussian approaches are either Frame-Stream, offering scalability but poor temporal stability, or Clip, achieving local consistency at the cost of high memory and limited sequence length. We propose ClipGStream, a hybrid reconstruction framework that performs stream optimization at the clip level rather than the frame level. The sequence is divided into short clips, where dynamic motion is modeled using clip-independent spatio-temporal fields and residual anchor compensation to capture local variations efficiently, while inter-clip inherited anchors and decoders maintain structural consistency across clips. This Clip-Stream design enables scalable, flicker-free reconstruction of long dynamic videos with high temporal coherence and reduced memory overhead. Extensive experiments demonstrate that ClipGStream achieves state-of-the-art reconstruction quality and efficiency. The project page is available at: https://liangjie1999.github.io/ClipGStreamWeb/

cs.CV↗
arXiv2026

SpikingMamba: Towards Energy-Efficient Large Language Models via Knowledge Distillation from Mamba

Large Language Models (LLMs) have achieved remarkable performance across tasks but remain energy-intensive due to dense matrix operations. Spiking neural networks (SNNs) improve energy efficiency by replacing dense matrix multiplications with sparse accumulations. Their sparse spike activity enables efficient LLMs deployment on edge devices. However, prior SNN-based LLMs often sacrifice performance for efficiency, and recovering accuracy typically requires full pretraining, which is costly and impractical. To address this, we propose SpikingMamba, an energy-efficient SNN-based LLMs distilled from Mamba that improves energy efficiency with minimal accuracy sacrifice. SpikingMamba integrates two key components: (a) SI-LIF, a signed-integer spiking neuron that preserves semantic polarity through signed multi-level spike representations. (b) A training-exclusive Smoothed Gradient Compensation (SGC) path mitigating quantization loss while preserving spike-driven efficiency. We employ a single-stage distillation strategy to transfer the zero-shot ability of pretrained Mamba and further enhance it via reinforcement learning (RL). Experiments show that SpikingMamba-1.3B achieves a 4.76$\times$ energy benefit, with only a 4.78\% zero-shot accuracy gap compared to the original Mamba. The model achieves a further 2.55\% accuracy improvement after RL, narrowing the performance gap from 4.78\% to 2.23\%. Code is available at: https://github.com/HuuYuLong/SpikingMamba .

cs.NE↗
arXiv2026

A Rigorous Proof of a Ramanujan Machine Identity for $-π/4$ via Exact Recurrence Solving

We prove a polynomial continued fraction identity for the constant $-π/4$, conjectured by the Ramanujan Machine project. The proof proceeds by explicitly solving the underlying second-order linear difference equation. We derive a closed-form expression for the denominator sequence, $q_n = (-1)^n (2n-3)!!\,(n^2+n-1)$, and establish absolute convergence via a Wronskian telescoping argument. The limiting value is reduced by Abel summation to a Beta-function integral, which is evaluated in closed form through an elementary substitution and a single integration by parts, yielding the exact value $-π/4$.

math.GM↗
arXiv2026

The Navier-Stokes equations in $\mathbb R^2_+$ with point vortex initial data: construction of the solution

This is the first of two papers concerning the asymptotic behavior of the incompressible Navier-Stokes equations in a half-space at high Reynolds numbers, with initial data given by a point vortex. In the present work, we establish the existence and uniqueness of solutions subject to the non-slip boundary condition. This result was established in \cite{Ken} under the condition that the total mass is sufficiently small. Here, we eliminate the smallness assumption by analyzing the linearized operator near the point vortex and constructing a tailored functional framework-one designed to capture the distinct behaviors of the solution in the vicinity of the point vortex and the boundary, respectively.

math.AP↗
arXiv2026

Probabilistic Evolution of Black Hole Thermodynamic States via Fokker-Planck Equation

Employing the generalized free energy landscape and solving the associated Fokker-Planck equation, we obtain the time-dependent probability evolution of the order parameter for the RN-AdS black hole phase transitions. Our analysis reveals two distinct kinetic regimes, namely relaxation dynamics initialized at the unstable maximum and phase transition from the metastable state. Furthermore, we characterize the non-equilibrium irreversibility and macroscopic uncertainty using the entropy production rate and the Shannon entropy. The results demonstrate that the phase transition synchronizes exactly with a prominent peak in the entropy production rate, identifying the barrier crossing event as a process fundamentally driven by maximum thermodynamic dissipation.

gr-qc↗
arXiv2026

The Navier-Stokes equations in $\mathbb R^2_+$ with point vortex initial data: Zero-viscosity limit

This is the second of two papers devoted to the asymptotic behavior of solutions to the incompressible Navier-Stokes equations in a half-space with point vortex initial data. A major difficulty stems from the interaction between the point vortex initial data and the boundary, which complicates the derivation of a valid asymptotic expansion. To overcome this, we carry out a precise matching between the point vortex and boundary-layer profiles to accurately capture the correct viscous behavior of the vortex in the half-plane. Based on this matched asymptotic analysis, we decompose the vorticity into three components: vorticity near the point vortex, vorticity near the boundary, and vorticity in the transition layer. A key point is that each component must be analyzed in its own distinct region. On this basis, we establish refined estimates and thereby achieve the inviscid limit for the point vortex. Finally, we rigorously prove that solutions to the Navier-Stokes equations converge to the Lamb-Oseen vortex away from the boundary, while approaching the Prandtl boundary-layer system in the near-boundary region.

math.AP↗