arXiv subjects

Li Wang

Publications and source records attributed to Li Wang.

At least 145 records · Page 8Linked to original sources

arXiv2025

ObCLIP: Oblivious CLoud-Device Hybrid Image Generation with Privacy Preservation

Diffusion Models have gained significant popularity due to their remarkable capabilities in image generation, albeit at the cost of intensive computation requirement. Meanwhile, despite their widespread deployment in inference services such as Midjourney, concerns about the potential leakage of sensitive information in uploaded user prompts have arisen. Existing solutions either lack rigorous privacy guarantees or fail to strike an effective balance between utility and efficiency. To bridge this gap, we propose ObCLIP, a plug-and-play safeguard that enables oblivious cloud-device hybrid generation. By oblivious, each input prompt is transformed into a set of semantically similar candidate prompts that differ only in sensitive attributes (e.g., gender, ethnicity). The cloud server processes all candidate prompts without knowing which one is the real one, thus preventing any prompt leakage. To mitigate server cost, only a small portion of denoising steps is performed upon the large cloud model. The intermediate latents are then sent back to the client, which selects the targeted latent and completes the remaining denoising using a small device model. Additionally, we analyze and incorporate several cache-based accelerations that leverage temporal and batch redundancy, effectively reducing computation cost with minimal utility degradation. Extensive experiments across multiple datasets demonstrate that ObCLIP provides rigorous privacy and comparable utility to cloud models with slightly increased server cost.

cs.CR↗

arXiv2025

NCV: A Node-Wise Consistency Verification Approach for Low-Cost Structured Error Localization in LLM Reasoning

Verifying multi-step reasoning in large language models is difficult due to imprecise error localization and high token costs. Existing methods either assess entire reasoning chains, suffering attention dilution, or rely on expensive multi-sampling. We introduce Node-wise Consistency Verification (NCV), a training-free framework that recasts verification as lightweight binary consistency checks at the node level. By decomposing the chain of thought into interconnected verification nodes, NCV precisely localizes errors and avoids unnecessary long-form generation. Experiments demonstrate that our approach enhances interpretability and efficiency, presenting a scalable solution for reliable LLM reasoning verification. On public datasets, NCV achieves a 10\% to 25\% improvement in F1 scores over baselines while utilizing $6\times$~$58\times$ fewer tokens than traditional methods like CoT-based verifiers.

cs.AI↗

arXiv2025

The first Brauer-Thrall conjecture for extriangulated length categories

Let $(\mathcal{A},Θ)$ be a length category. We introduce the notation of Gabriel-Roiter measure with respect to $Θ$ and extend Gabriel's main property to this setting. Using this measure, when $(\mathcal{A},Θ)$ satisfies some technical conditions, we prove that $\mathcal{A}$ has an infinite number of pairwise nonisomorphic indecomposable objects if and only if it has indecomposable objects of arbitrarily large length. That is, the first Brauer-Thrall conjecture holds.

math.RT↗

arXiv2025

An Investigation of Batch Normalization in Off-Policy Actor-Critic Algorithms

Batch Normalization (BN) has played a pivotal role in the success of deep learning by improving training stability, mitigating overfitting, and enabling more effective optimization. However, its adoption in deep reinforcement learning (DRL) has been limited due to the inherent non-i.i.d. nature of data and the dynamically shifting distributions induced by the agent's learning process. In this paper, we argue that, despite these challenges, BN retains unique advantages in DRL settings, particularly through its stochasticity and its ability to ease training. When applied appropriately, BN can adapt to evolving data distributions and enhance both convergence speed and final performance. To this end, we conduct a comprehensive empirical study on the use of BN in off-policy actor-critic algorithms, systematically analyzing how different training and evaluation modes impact performance. We further identify failure modes that lead to instability or divergence, analyze their underlying causes, and propose the Mode-Aware Batch Normalization (MA-BN) method with practical actionable recommendations for robust BN integration in DRL pipelines. We also empirically validate that, in RL settings, MA-BN accelerates and stabilizes training, broadens the effective learning rate range, enhances exploration, and reduces overall optimization difficulty. Our code is available at: https://github.com/monster476/ma-bn.git.

cs.LG↗

arXiv2025

Entanglement manifestation of knot topology in a non-Hermitian lattice

Although the homotopy-knot theory has been utilized to implement effective topological classification for non-Hermitian systems, the physical implications underlying distinct knot topologies remain ambiguous and are rarely addressed. In this work, we propose a one-dimensional non-Hermitian four-band lattice model and map out its phase diagram according to the distinct knot structures residing in the moment space. The topological phase diagram is ascertained through a spectral winding number. Furthermore, we derive the exact analytic formula for the phase boundaries that delineate different knot topologies. To explore the concrete physical implications of distinct knot topologies, we investigate the many-body ground state entanglement entropy for free fermions loaded on such non-Hermitian lattice in real space. It turns out that different knot topologies imply different magnitudes of entanglement. Moreover, we show that the central charge c extracted from systematic finite-size scaling of entanglement entropy provides effective description of the phase diagram of the knot topology. Finally, we further confirm the phase boundaries for the topological phase transitions alternatively by numerical calculations of the many-body ground state fidelity susceptibility. Our results showcase the connection between knot topology and entanglement of non-Hermitian systems and may facilitate further exploration of the profound and practical physical implications of knot topology.

quant-ph↗

arXiv2025

Over-the-Air Adversarial Attack Detection: from Datasets to Defenses

Automatic Speaker Verification (ASV) systems can be used for voice-enabled applications for identity verification. However, recent studies have exposed these systems' vulnerabilities to both over-the-line (OTL) and over-the-air (OTA) adversarial attacks. Although various detection methods have been proposed to counter these threats, they have not been thoroughly tested due to the lack of a comprehensive data set. To address this gap, we developed the AdvSV 2.0 dataset, which contains 628k samples with a total duration of 800 hours. This dataset incorporates classical adversarial attack algorithms, ASV systems, and encompasses both OTL and OTA scenarios. Furthermore, we introduce a novel adversarial attack method based on a Neural Replay Simulator (NRS), which enhances the potency of adversarial OTA attacks, thereby presenting a greater threat to ASV systems. To defend against these attacks, we propose CODA-OCC, a contrastive learning approach within the one-class classification framework. Experimental results show that CODA-OCC achieves an EER of 11.2% and an AUC of 0.95 on the AdvSV 2.0 dataset, outperforming several state-of-the-art detection methods.

eess.AS↗

arXiv2025

Audio Deepfake Verification

With the rapid development of deepfake technology, simply making a binary judgment of true or false on audio is no longer sufficient to meet practical needs. Accurately determining the specific deepfake method has become crucial. This paper introduces the Audio Deepfake Verification (ADV) task, effectively addressing the limitations of existing deepfake source tracing methods in closed-set scenarios, aiming to achieve open-set deepfake source tracing. Meanwhile, the Audity dual-branch architecture is proposed, extracting deepfake features from two dimensions: audio structure and generation artifacts. Experimental results show that the dual-branch Audity architecture outperforms any single-branch configuration, and it can simultaneously achieve excellent performance in both deepfake detection and verification tasks.

eess.AS↗

arXiv2025

MEGG: Replay via Maximally Extreme GGscore in Incremental Learning for Neural Recommendation Models

Neural Collaborative Filtering models are widely used in recommender systems but are typically trained under static settings, assuming fixed data distributions. This limits their applicability in dynamic environments where user preferences evolve. Incremental learning offers a promising solution, yet conventional methods from computer vision or NLP face challenges in recommendation tasks due to data sparsity and distinct task paradigms. Existing approaches for neural recommenders remain limited and often lack generalizability. To address this, we propose MEGG, Replay Samples with Maximally Extreme GGscore, an experience replay based incremental learning framework. MEGG introduces GGscore, a novel metric that quantifies sample influence, enabling the selective replay of highly influential samples to mitigate catastrophic forgetting. Being model-agnostic, MEGG integrates seamlessly across architectures and frameworks. Experiments on three neural models and four benchmark datasets show superior performance over state-of-the-art baselines, with strong scalability, efficiency, and robustness. Implementation will be released publicly upon acceptance.

cs.IR↗

arXiv2025

TeRA: Rethinking Text-guided Realistic 3D Avatar Generation

In this paper, we rethink text-to-avatar generative models by proposing TeRA, a more efficient and effective framework than the previous SDS-based models and general large 3D generative models. Our approach employs a two-stage training strategy for learning a native 3D avatar generative model. Initially, we distill a decoder to derive a structured latent space from a large human reconstruction model. Subsequently, a text-controlled latent diffusion model is trained to generate photorealistic 3D human avatars within this latent space. TeRA enhances the model performance by eliminating slow iterative optimization and enables text-based partial customization through a structured 3D human representation. Experiments have proven our approach's superiority over previous text-to-avatar generative models in subjective and objective evaluation.

cs.CV↗

arXiv2025

A magnetic white dwarf formed through a binary merger within 35 million years

White dwarfs (WDs) represent the final evolutionary stage of most stars, typically originating from progenitor stars with masses below approximately 8 $M_{\odot}$ to 10 $M_{\odot}$. Formation through single-star evolution generally requires at least 25 Myr, with the youngest WDs often near the Chandrasekhar limit of 1.4 $M_{\odot}$. In contrast, WDs formed via binary channels, such as mergers or mass transfer, can develop smaller masses in a shorter timescale and may exhibit unique characteristics, including strong surface magnetic fields and rapid rotation. Accurately determining the ages of these WDs is essential for understanding their formation. A valuable method involves studying WDs in star clusters, where member stars share the same age and chemical composition, allowing for precise constraints on the formation times and metallicities of the WDs' progenitors. Here we report a WD found in the open cluster RSG 5, which is only 35 Myr old. The WD's mass is lower than 1.05 $M_{\odot}$, indicating it may not have formed through single-star evolution. The WD possesses an exceptionally strong surface magnetic field ($\ge 200$ MG), a short rotational period ($\sim 6.5$ min), and, most notably, a co-rotating half-ring of ionized circumstellar debris. This distinctive feature provides evidence for a binary merger origin, a scenario further substantiated by our stellar evolution models.

astro-ph.SR↗

arXiv2025

Noro: Noise-Robust One-shot Voice Conversion with Hidden Speaker Representation Learning

The effectiveness of one-shot voice conversion (VC) decreases in real-world scenarios where reference speeches, which are often sourced from the internet, contain various disturbances like background noise. To address this issue, we introduce Noro, a noise-robust one-shot VC system. Noro features innovative components tailored for VC using noisy reference speeches, including a dual-branch reference encoding module and a noise-agnostic contrastive speaker loss. Experimental results demonstrate that Noro outperforms our baseline system in both clean and noisy scenarios, highlighting its efficacy for real-world applications. Additionally, we investigate the hidden speaker representation capabilities of our baseline system by repurposing its reference encoder as a speaker encoder. The results show that it is competitive with several advanced self-supervised learning models for speaker representation under the SUPERB settings, highlighting the potential for advancing speaker representation learning through one-shot VC tasks.

cs.SD↗

arXiv2025

Controlling instability in the Vlasov-Poisson system through moment-based optimization

Controlling instability in plasma is one of the central challenges in fusion energy research. Among the various sources of instability, kinetic effects play a significant role. In this work, we aim to suppress the instability induced by kinetic effects by designing an external electric field. However, rather than directly solving the full kinetic Vlasov-Poisson system, we focus on a reduced-order model, specifically the moment-based system, to capture the underlying dynamics. This approach is motivated by the desire to reduce the computational cost associated with repeatedly solving the high-dimensional kinetic equations during the optimization of the electric field. Additionally, moment-based data is more readily accessible in practice, making a moment-based control framework more adaptable to data-driven scenarios. We investigate the effectiveness of moment-based control both analytically and numerically, by comparing it to control based on the full kinetic model.

math.NA↗

arXiv2025

Autonomous Driving with RSMA-Enabled Finite Blocklength Transmissions: Ergodic Performance Analysis and Optimization

Rate-splitting multiple access (RSMA) is a key technology for next-generation multiple access systems due to its robustness against imperfect channel state information (CSI). This makes RSMA particularly suitable for high-mobility autonomous driving, where ultra-reliable and low-latency communication (URLLC) is essential. To address the stringent requirements, this study enables RSMA finite blocklength (FBL) transmissions and explicitly evaluates the ergodic performance. We derive the closed-form lower bound for the ergodic sum-rate of RSMA, considering vital factors such as the vehicle velocities, vehicle positions, power allocation of each stream, blocklengths, and block error rates (BLERs). To further enhance the ergodic sum-rate while complying with quality of service (QoS) rate constraints, we jointly optimize the global power coefficient, private power distribution, and common rate splitting. Guided by gradient descent, we first adjust the global power coefficient based on its sum-rate solution. This parameter regulates the power state of the common stream, allowing for dynamic activation or deactivation: if active, we optimize the private power distribution and adjust the common rate splitting to meet minimum transmission constraints; if inactive, we use the sequential quadratic programming for private power distribution optimization. Simulation results confirm that our RSMA scheme significantly improves the ergodic performance, reduces blocklength and BLER, surpassing the RSMA counterpart with average private power and space division multiple access (SDMA). Furthermore, our approach is validated to guarantee the rates for users with the poorest channel conditions, thereby enhancing fairness across the network.

eess.SP↗

arXiv2025

KG-Augmented Executable CoT for Mathematical Coding

In recent years, large language models (LLMs) have excelled in natural language processing tasks but face significant challenges in complex reasoning tasks such as mathematical reasoning and code generation. To address these limitations, we propose KG-Augmented Executable Chain-of-Thought (KGA-ECoT), a novel framework that enhances code generation through knowledge graphs and improves mathematical reasoning via executable code. KGA-ECoT decomposes problems into a Structured Task Graph, leverages efficient GraphRAG for precise knowledge retrieval from mathematical libraries, and generates verifiable code to ensure computational accuracy. Evaluations on multiple mathematical reasoning benchmarks demonstrate that KGA-ECoT significantly outperforms existing prompting methods, achieving absolute accuracy improvements ranging from several to over ten percentage points. Further analysis confirms the critical roles of GraphRAG in enhancing code quality and external code execution in ensuring precision. These findings collectively establish KGA-ECoT as a robust and highly generalizable framework for complex mathematical reasoning tasks.

cs.AI↗

arXiv2025

Don't Overthink It: A Survey of Efficient R1-style Large Reasoning Models

Recently, Large Reasoning Models (LRMs) have gradually become a research hotspot due to their outstanding performance in handling complex tasks. Among them, DeepSeek R1 has garnered significant attention for its exceptional performance and open-source nature, driving advancements in the research of R1-style LRMs. Unlike traditional Large Language Models (LLMs), these models enhance logical deduction and decision-making capabilities during reasoning by incorporating mechanisms such as long chain-of-thought and self-reflection through reinforcement learning. However, with the widespread application of these models, the problem of overthinking has gradually emerged. Specifically, when generating answers, these models often construct excessively long reasoning chains with redundant or repetitive steps, which leads to reduced reasoning efficiency and may affect the accuracy of the final answer. To this end, various efficient reasoning methods have been proposed, aiming to reduce the length of reasoning paths without compromising model performance and reasoning capability. By reviewing the current research advancements in the field of efficient reasoning methods systematically, we categorize existing works into two main directions based on the lens of single-model optimization versus model collaboration: (1) Efficient Reasoning with Single Model, which focuses on improving the reasoning efficiency of individual models; and (2) Efficient Reasoning with Model Collaboration, which explores optimizing reasoning paths through collaboration among multiple models. Besides, we maintain a public GitHub repository that tracks the latest progress in efficient reasoning methods.

cs.AI↗

arXiv2025

Dynamic borrowing from historical controls via the synthetic prior with covariates in randomized clinical trials

Motivated by a rheumatoid arthritis clinical trial, we propose a new Bayesian method called SPx, standing for synthetic prior with covariates, to borrow information from historical trials to reduce the control group size in a new trial. The method involves a novel use of Bayesian model averaging to balance between multiple possible relationships between the historical and new trial data, allowing the historical data to be dynamically trusted or discounted as appropriate. We require only trial-level summary statistics, which are available more often than patient-level data. Through simulations and an application to the rheumatoid arthritis trial we show that SPx can substantially reduce the control group size while maintaining Frequentist properties.

stat.ME↗

arXiv2025

Continuous data assimilation for hydrodynamics: consistent discretization and application to moment recovery

Motivated by the challenge of moment recovery in hydrodynamic approximation in kinetic theory, we propose a data-driven approach for the hydrodynamic models. Inspired by continuous data assimilation, our method introduces a relaxation-based nudging system coupled with a novel discretization technique. This approach facilitates the simultaneous recovery of both the force term and a high-resolution solution from sparsely observed data. To address potential numerical artifacts, we use kernel regression to fit the observed data. We also analyze the convergence of the proposed nudging system under both full and partial data scenarios. When applied to moment systems, the source term involves the derivative of higher-order moments, our approach serves as a crucial step for data preparation in machine-learning based moment closure models. Multiple numerical experiments demonstrate the effectiveness of our algorithm, and we discuss its potential extension to high-dimensional systems.

math.NA↗

arXiv2025

Pattern formations of coupled PDEs with transparent boundary conditions in product-type ends and applications

This paper studies pattern formations in coupled elliptic PDE systems governed by transparent boundary conditions. Such systems unify diverse areas, including inverse boundary problems (via a single passive/active boundary measurement), spectral geometry of transmission eigenfunctions, and geometric characterization of invisibility phenomena and inverse shape problems in wave scattering. We uncover and rigorously characterize a novel local pattern formation, establishing a sharp quantitative relationship between the difference in the PDEs' lower-order terms and the geometric/regularity parameters within a generic domain's product-type ends-structures characterized by high extrinsic curvature. This foundational result yields new findings with novel physical insights and practical implications across these fields.

math.AP↗