arXiv subjects

Lei Wang

Publications and source records attributed to Lei Wang.

At least 433 records · Page 24Linked to original sources

arXiv2025

A Survey on Test-Time Scaling in Large Language Models: What, How, Where, and How Well?

As enthusiasm for scaling computation (data and parameters) in the pretraining era gradually diminished, test-time scaling (TTS), also referred to as ``test-time computing'' has emerged as a prominent research focus. Recent studies demonstrate that TTS can further elicit the problem-solving capabilities of large language models (LLMs), enabling significant breakthroughs not only in specialized reasoning tasks, such as mathematics and coding, but also in general tasks like open-ended Q&A. However, despite the explosion of recent efforts in this area, there remains an urgent need for a comprehensive survey offering a systemic understanding. To fill this gap, we propose a unified, multidimensional framework structured along four core dimensions of TTS research: what to scale, how to scale, where to scale, and how well to scale. Building upon this taxonomy, we conduct an extensive review of methods, application scenarios, and assessment aspects, and present an organized decomposition that highlights the unique functional roles of individual techniques within the broader TTS landscape. From this analysis, we distill the major developmental trajectories of TTS to date and offer hands-on guidelines for practical deployment. Furthermore, we identify several open challenges and offer insights into promising future directions, including further scaling, clarifying the functional essence of techniques, generalizing to more tasks, and more attributions. Our repository is available on https://github.com/testtimescaling/testtimescaling.github.io/

cs.CL↗

arXiv2025

SignLLM: Sign Language Production Large Language Models

In this paper, we propose SignLLM, a multilingual Sign Language Production (SLP) large language model, which includes two novel multilingual SLP modes MLSF and Prompt2LangGloss that allow sign language gestures generation from query texts input and question-style prompts input respectively. Both modes can use a new RL loss based on reinforcement learning and a new RL module named Priority Learning Channel. These RL components can accelerate the training by enhancing the model's capability to sample high-quality data. To train SignLLM, we introduce Prompt2Sign, a comprehensive multilingual sign language dataset, which builds from public data, including American Sign Language (ASL) and seven others. This dataset standardizes information by extracting pose information from sign language videos into a unified compressed format. We extensively evaluate SignLLM, demonstrating that our model achieves state-of-the-art performance on SLP tasks across eight sign languages.

cs.CV↗

arXiv2025

A Unified Variational Framework for Quantum Excited States

Determining quantum excited states is crucial across physics and chemistry but presents significant challenges for variational methods, primarily due to the need to enforce orthogonality to lower-energy states, often requiring state-specific optimization, penalty terms, or specialized ansatz constructions. We introduce a novel variational principle that overcomes these limitations, enabling the \textit{simultaneous} determination of multiple low-energy excited states. The principle is based on minimizing the trace of the inverse overlap matrix multiplied by the Hamiltonian matrix, $\mathrm{Tr}(\mathbf{S}^{-1}\mathbf{H})$, constructed from a set of \textit{non-orthogonal} variational states $\{|ψ_i\rangle\}$. Here, $\mathbf{H}_{ij} = \langleψ_i | H | ψ_j\rangle$ and $\mathbf{S}_{ij} = \langleψ_i | ψ_j\rangle$ are the elements of the Hamiltonian and overlap matrices, respectively. This approach variationally optimizes the entire low-energy subspace spanned by $\{|ψ_i\rangle\}$ without explicit orthogonality constraints or penalty functions. We demonstrate the power and generality of this method across diverse physical systems and variational ansatzes: calculating the low-energy spectrum of 1D Heisenberg spin chains using matrix product states, finding vibrational spectrum of Morse potential using quantics tensor trains for real-space wavefunctions, and determining excited states for 2D fermionic Hubbard model with variational quantum circuits. In all applications, the method accurately and simultaneously obtains multiple lowest-lying energy levels and their corresponding states, showcasing its potential as a unified and flexible framework for calculating excited states on both classical and quantum computational platforms.

quant-ph↗

arXiv2025

Materials discovery acceleration by using condition generative methodology

With the rapid advancement of AI technologies, generative models have been increasingly employed in the exploration of novel materials. By integrating traditional computational approaches such as density functional theory (DFT) and molecular dynamics (MD), existing generative models, including diffusion models and autoregressive models, have demonstrated remarkable potential in the discovery of novel materials. However, their efficiency in goal-directed materials design remains suboptimal. In this work we developed a highly transferable, efficient and robust conditional generation framework, PODGen, by integrating a general generative model with multiple property prediction models. Based on PODGen, we designed a workflow for the high-throughput crystals conditional generation which is used to search new topological insulators (TIs). Our results show that the success rate of generating TIs using our framework is 5.3 times higher than that of the unconstrained approach. More importantly, while general methods rarely produce gapped TIs, our framework succeeds consistently, highlighting an effectively $\infty$ improvement. This demonstrates that conditional generation significantly enhances the efficiency of targeted material discovery. Using this method, we generated tens of thousands of new topological materials and conducted further first-principles calculations on those with promising application potential. Furthermore, we identified promising, synthesizable topological (crystalline) insulators such as CsHgSb, NaLaB$_{12}$, Bi$_4$Sb$_2$Se$_3$, Be$_3$Ta$_2$Si and Be$_2$W.

cond-mat.mtrl-sci↗

arXiv2025

Region of Synchronization Estimation for Complex Networks via SOS Programming

In this article, we explore the problem of the region of synchronization (ROS) for complex networks with nonlinear dynamics. Given a pair of state- and target- sets, our goal is to estimate the ROS such that the trajectories originating within it reach the target set (i.e., synchronization manifold), without leaving the state set before the first hitting time. In order to do so, an exponential guidance-barrier function is proposed to construct the ROS along the synchronization manifold, and the corresponding sufficient conditions for estimating the ROS are developed. The resulting conditions lead to a sum-of-squares programming problem, thereby affording a polynomial-time solvability. Furthermore, when the synchronization manifold reduces to an equilibrium point, our method not only estimates a larger ROS compared to existing results but also allows the ROS to take more general shapes. Finally, we present two numerical examples to demonstrate the effectiveness of the theoretical results.

math.OC↗

arXiv2025

A Survey on Parameter-Efficient Fine-Tuning for Foundation Models in Federated Learning

Foundation models have revolutionized artificial intelligence by providing robust, versatile architectures pre-trained on large-scale datasets. However, adapting these massive models to specific downstream tasks requires fine-tuning, which can be prohibitively expensive in computational resources. Parameter-Efficient Fine-Tuning (PEFT) methods address this challenge by selectively updating only a small subset of parameters. Meanwhile, Federated Learning (FL) enables collaborative model training across distributed clients without sharing raw data, making it ideal for privacy-sensitive applications. This survey provides a comprehensive review of the integration of PEFT techniques within federated learning environments. We systematically categorize existing approaches into three main groups: Additive PEFT (which introduces new trainable parameters), Selective PEFT (which fine-tunes only subsets of existing parameters), and Reparameterized PEFT (which transforms model architectures to enable efficient updates). For each category, we analyze how these methods address the unique challenges of federated settings, including data heterogeneity, communication efficiency, computational constraints, and privacy concerns. We further organize the literature based on application domains, covering both natural language processing and computer vision tasks. Finally, we discuss promising research directions, including scaling to larger foundation models, theoretical analysis of federated PEFT methods, and sustainable approaches for resource-constrained environments.

cs.LG↗

arXiv2025

Two-dimensional Dual-Switchable Ferroelectric Altermagnets: Altering Electrons and Magnons

Ferroelectric altermagnets (FEAMs) offer unique magnetoelectric coupling properties by combining the characteristics of both antiferromagnets and ferromagnets, yet their multifunctional electric control remains largely unexplored. Here, we introduce and investigate a scenario for the simultaneous electrical switching of electronic spin and magnonic chirality splitting in two-dimensional FEAMs. Based on the C2DB database, employing symmetry analysis and first-principles calculations, we study prototypical candidates CrPS$_3$ and V$_2$I$_2$O$_2$BrCl. We identify the mechanism: ferroelectricity arises from asymmetric displacements (P along $z$ in CrPS$_3$, V along the $xy$-direction in V$_2$I$_2$O$_2$BrCl), which inherently couples electric polarization to both electronic and magnonic degrees of freedom by retaining [C$_2$$||$M] symmetry. Our calculations explicitly demonstrate that reversing the ferroelectric polarization concurrently switches the sign of the electronic spin splitting and chirality of magnonic modes. This shows these materials as dual-switchable FEAMs, enabling unified electrical manipulation of electron and magnon properties. A potentially experimentally detectable method via the magneto-optical Kerr effect was derived. This work provides a materials-specific realization and theoretical basis for designing novel electrically controlled multifunctional spintronic, spin caloritronic, and magnonic devices.

cond-mat.mtrl-sci↗

arXiv2025

TileLang: A Composable Tiled Programming Model for AI Systems

Modern AI workloads rely heavily on optimized computing kernels for both training and inference. These AI kernels follow well-defined data-flow patterns, such as moving tiles between DRAM and SRAM and performing a sequence of computations on those tiles. However, writing high-performance kernels remains complex despite the clarity of these patterns. Achieving peak performance requires careful, hardware-centric optimizations to fully leverage modern accelerators. While domain-specific compilers attempt to reduce the burden of writing high-performance kernels, they often struggle with usability and expressiveness gaps. In this paper, we present TileLang, a generalized tiled programming model for more efficient AI Kernel programming. TileLang decouples scheduling space (thread binding, layout, tensorize and pipeline) from dataflow, and encapsulated them as a set of customization annotations and primitives. This approach allows users to focus on the kernel's data-flow itself, while leaving most other optimizations to compilers. We conduct comprehensive experiments on commonly-used devices, across numerous experiments, our evaluation shows that TileLang can achieve state-of-the-art performance in key kernels, demonstrating that its unified block-and-thread paradigm and transparent scheduling capabilities deliver both the power and flexibility demanded by modern AI system development.

cs.LG↗

arXiv2025

ChiseLLM: Unleashing the Power of Reasoning LLMs for Chisel Agile Hardware Development

The growing demand for Domain-Specific Architecture (DSA) has driven the development of Agile Hardware Development Methodology (AHDM). Hardware Construction Language (HCL) like Chisel offers high-level abstraction features, making it an ideal language for HCL-Based AHDM. While Large Language Models (LLMs) excel in code generation tasks, they still face challenges with Chisel generation, particularly regarding syntax correctness and design variability. Recent reasoning models have significantly enhanced code generation capabilities through test-time scaling techniques. However, we found that reasoning models without domain adaptation cannot bring substantial benefits to Chisel code generation tasks. This paper presents ChiseLLM, a solution comprising data processing and transformation, prompt-guided reasoning trace synthesis, and domain-adapted model training. We constructed high-quality datasets from public RTL code resources and guided the model to adopt structured thinking patterns through prompt enhancement methods. Experiments demonstrate that our ChiseLLM-7B and ChiseLLM-32B models improved syntax correctness by 18.85% and 26.32% respectively over base models, while increasing variability design ability by 47.58% compared to baseline reasoning models. Our datasets and models are publicly available, providing high-performance, cost-effective models for HCL-Based AHDM, and offering an effective baseline for future research. Github repository: https://github.com/observerw/ChiseLLM

cs.AI↗

arXiv2025

Discovery of a Low-mass Strong-lens System in SMACS J0723.3-7327

We report the discovery of an intriguing, low-mass galaxy-scale strong-lens system in the SMACS J0723.3-7327 galaxy cluster. By modeling James Webb Space Telescope imaging and Very Large Telescope Multi-Unit Spectroscopic Explorer spectroscopic data, we find that the lens is cluster member galaxy at $z=0.397$ with an Einstein radius of $0^{\prime \prime}.424$ $\pm$ $0^{\prime \prime}.012$, stellar mass of $M_* = (3.3 \pm 0.8) \times 10^{10} M_\odot$, half-light radius of $\sim 1$ kpc, and central stellar velocity dispersion of $140 \pm 6$ km s$^{-1}$. This lens galaxy is one of the few strong lens galaxies known to date that have stellar mass as low as $M_* \sim 10^{10.5} M_\odot$, offering an exceptional opportunity to peek into the population of low-mass galaxies that has largely remained unexplored in the context of strong-lensing studies. This strong lens system can also assist in assessing the systematic uncertainty in the lens modeling of cluster member galaxies.

astro-ph.GA↗

arXiv2025

Embodied Neuromorphic Control Applied on a 7-DOF Robotic Manipulator

The development of artificial intelligence towards real-time interaction with the environment is a key aspect of embodied intelligence and robotics. Inverse dynamics is a fundamental robotics problem, which maps from joint space to torque space of robotic systems. Traditional methods for solving it rely on direct physical modeling of robots which is difficult or even impossible due to nonlinearity and external disturbance. Recently, data-based model-learning algorithms are adopted to address this issue. However, they often require manual parameter tuning and high computational costs. Neuromorphic computing is inherently suitable to process spatiotemporal features in robot motion control at extremely low costs. However, current research is still in its infancy: existing works control only low-degree-of-freedom systems and lack performance quantification and comparison. In this paper, we propose a neuromorphic control framework to control 7 degree-of-freedom robotic manipulators. We use Spiking Neural Network to leverage the spatiotemporal continuity of the motion data to improve control accuracy, and eliminate manual parameters tuning. We validated the algorithm on two robotic platforms, which reduces torque prediction error by at least 60% and performs a target position tracking task successfully. This work advances embodied neuromorphic control by one step forward from proof of concept to applications in complex real-world tasks.

cs.RO↗

arXiv2025

Spike-Kal: A Spiking Neuron Network Assisted Kalman Filter

Kalman filtering can provide an optimal estimation of the system state from noisy observation data. This algorithm's performance depends on the accuracy of system modeling and noise statistical characteristics, which are usually challenging to obtain in practical applications. The powerful nonlinear modeling capabilities of deep learning, combined with its ability to extract features from large amounts of data automatically, offer new opportunities for improving the Kalman filter. This paper proposes a novel method that leverages the Spiking Neural Network to optimize the Kalman filter. Our approach aims to reduce the reliance on prior knowledge of system and observation noises, allowing for adaptation to varying statistical characteristics of time-varying noise. Furthermore, we investigate the potential of SNNs in improving the computational efficiency of the Kalman filter. In our method, we design an integration strategy between the SNN and the Kalman filter. The SNN is trained to directly approximate the optimal gain matrix from observation data, thereby alleviating the computational burden of complex matrix operations inherent in traditional Kalman filtering while maintaining the accuracy and robustness of state estimation. Its average error has been reduced by 18\%-65\% compared with other methods.

eess.SY↗

arXiv2025

Phenomenological study of the angle between jet axes in heavy-ion collisions

This paper presents a phenomenological study on the angle between the Standard and the Winner-Take-All (WTA) jet axes ($ΔR_{\rm axis}^{\rm WTA-Std}$) in high-energy nuclear collisions. The $p$+$p$ baseline is provided by the Pythia8 event generator. The in-medium jet propagation is simulated by the linear Boltzmann transport (LBT) model, which considers both the elastic and inelastic jet-medium interactions. Our theoretical results calculated by the LBT model show that the $ΔR_{\rm axis}^{\rm WTA-Std}$ distribution in Pb+Pb at $\sqrt{s}=5.02$ TeV is narrower than that in $p$+$p$, which agrees well with the recent ALICE measurements. The narrowing of $ΔR_{\rm axis}^{\rm WTA-Std}$ seems to violate the $p_T$-broadening nature of the jet quenching effect, usually explained by the influence of "selection bias". However, the physical details still need to be fully understood. Utilizing a matching-jet method to track the jet evolution in the QGP to remove the selection bias in the Monte Carlo simulations, we observe that the $ΔR_{\rm axis}^{\rm WTA-Std}$ distribution becomes broader due to the jet-medium interactions. At the same time, by rescaling the quark/gluon-jet fractions in Pb+Pb collisions to be the same as that in $p$+$p$, we find that the fraction change may not significantly influence the modification pattern of jet $ΔR_{\rm axis}^{\rm WTA-Std}$. On the other hand, the selected jet sample in A+A collisions has a significantly narrower initial $ΔR_{\rm axis}^{\rm WTA-Std}$ distribution than the $p$+$p$ baseline, and such a biased comparison between $p$+$p$ and A+A conceals the actual jet-broadening effect in the experimental measurements. The investigations presented in this paper will deepen our understanding of the relationship between the actual intra-jet modifications in the QGP and the experimental observations.

hep-ph↗

arXiv2025

Weakly-Supervised Learning via Multi-Lateral Decoder Branching for Tool Segmentation in Robot-Assisted Cardiovascular Catheterization

Robot-assisted catheterization has garnered a good attention for its potentials in treating cardiovascular diseases. However, advancing surgeon-robot collaboration still requires further research, particularly on task-specific automation. For instance, automated tool segmentation can assist surgeons in visualizing and tracking of endovascular tools during cardiac procedures. While learning-based models have demonstrated state-of-the-art segmentation performances, generating ground-truth labels for fully-supervised methods is both labor-intensive time consuming, and costly. In this study, we propose a weakly-supervised learning method with multi-lateral pseudo labeling for tool segmentation in cardiovascular angiogram datasets. The method utilizes a modified U-Net architecture featuring one encoder and multiple laterally branched decoders. The decoders generate diverse pseudo labels under different perturbations, augmenting available partial labels. The pseudo labels are self-generated using a mixed loss function with shared consistency across the decoders. The weakly-supervised model was trained end-to-end and validated using partially annotated angiogram data from three cardiovascular catheterization procedures. Validation results show that the model could perform closer to fully-supervised models. Also, the proposed weakly-supervised multi-lateral method outperforms three well known methods used for weakly-supervised learning, offering the highest segmentation performance across the three angiogram datasets. Furthermore, numerous ablation studies confirmed the model's consistent performance under different parameters. Finally, the model was applied for tool segmentation in a robot-assisted catheterization experiments. The model enhanced visualization with high connectivity indices for guidewire and catheter, and a mean processing time of 35 ms per frame.

cs.CV↗

arXiv2025

Experimental Implementation of a Qubit-Efficient Variational Quantum Eigensolver with Analog Error Mitigation on a Superconducting Quantum Processor

We experimentally demonstrate a qubit-efficient variational quantum eigensolver (VQE) algorithm using a superconducting quantum processor, employing minimal quantum resources with only a transmon qubit coupled to a high-coherence photonic qubit. By leveraging matrix product states to compress the quantum state representation, we simulate an N + 1-spin circular Ising model with a transverse field. Furthermore, we develop an analog error mitigation approach through zero-noise extrapolation by introducing a precise noise injection technique for the transmon qubit. As a validation, we apply our error-mitigated qubit-efficient VQE in determining the ground state energies of a 4-spin Ising model. Our results demonstrate the feasibility of performing quantum algorithms with minimal quantum resources while effectively mitigating the impact of noise, offering a promising pathway to bridge the gap between theoretical advances and practical implementations on current noisy intermediate-scale quantum devices.

quant-ph↗

arXiv2025

LeakGuard: Detecting Memory Leaks Accurately and Scalably

Memory leaks are prevalent in various real-world software projects, thereby leading to serious attacks like denial-of-service. Though prior methods for detecting memory leaks made significant advance, they often suffer from low accuracy and weak scalability for testing large and complex programs. In this paper we present LeakGuard, a memory leak detection tool which provides satisfactory balance of accuracy and scalability. For accuracy, LeakGuard analyzes the behaviors of library and developer-defined memory allocation and deallocation functions in a path-sensitive manner and generates function summaries for them in a bottom-up approach. Additionally, we develop a pointer escape analysis technique to model the transfer of pointer ownership. For scalability, LeakGuard examines each function of interest independently by using its function summary and under-constrained symbolic execution technique, which effectively mitigates path explosion problem. Our extensive evaluation on 18 real-world software projects and standard benchmark datasets demonstrates that LeakGuard achieves significant advancements in multiple aspects: it exhibits superior MAD function identification capability compared to Goshawk, outperforms five state-of-the-art methods in defect detection accuracy, and successfully identifies 129 previously undetected memory leak bugs, all of which have been independently verified and confirmed by the respective development teams.

cs.CR↗

arXiv2025

AP-CAP: Advancing High-Quality Data Synthesis for Animal Pose Estimation via a Controllable Image Generation Pipeline

The task of 2D animal pose estimation plays a crucial role in advancing deep learning applications in animal behavior analysis and ecological research. Despite notable progress in some existing approaches, our study reveals that the scarcity of high-quality datasets remains a significant bottleneck, limiting the full potential of current methods. To address this challenge, we propose a novel Controllable Image Generation Pipeline for synthesizing animal pose estimation data, termed AP-CAP. Within this pipeline, we introduce a Multi-Modal Animal Image Generation Model capable of producing images with expected poses. To enhance the quality and diversity of the generated data, we further propose three innovative strategies: (1) Modality-Fusion-Based Animal Image Synthesis Strategy to integrate multi-source appearance representations, (2) Pose-Adjustment-Based Animal Image Synthesis Strategy to dynamically capture diverse pose variations, and (3) Caption-Enhancement-Based Animal Image Synthesis Strategy to enrich visual semantic understanding. Leveraging the proposed model and strategies, we create the MPCH Dataset (Modality-Pose-Caption Hybrid), the first hybrid dataset that innovatively combines synthetic and real data, establishing the largest-scale multi-source heterogeneous benchmark repository for animal pose estimation to date. Extensive experiments demonstrate the superiority of our method in improving both the performance and generalization capability of animal pose estimators.

cs.CV↗

arXiv2025

Hierarchical Local-Global Feature Learning for Few-shot Malicious Traffic Detection

With the rapid growth of internet traffic, malicious network attacks have become increasingly frequent and sophisticated, posing significant threats to global cybersecurity. Traditional detection methods, including rule-based and machine learning-based approaches, struggle to accurately identify emerging threats, particularly in scenarios with limited samples. While recent advances in few-shot learning have partially addressed the data scarcity issue, existing methods still exhibit high false positive rates and lack the capability to effectively capture crucial local traffic patterns. In this paper, we propose HLoG, a novel hierarchical few-shot malicious traffic detection framework that leverages both local and global features extracted from network sessions. HLoG employs a sliding-window approach to segment sessions into phases, capturing fine-grained local interaction patterns through hierarchical bidirectional GRU encoding, while simultaneously modeling global contextual dependencies. We further design a session similarity assessment module that integrates local similarity with global self-attention-enhanced representations, achieving accurate and robust few-shot traffic classification. Comprehensive experiments on three meticulously reconstructed datasets demonstrate that HLoG significantly outperforms existing state-of-the-art methods. Particularly, HLoG achieves superior recall rates while substantially reducing false positives, highlighting its effectiveness and practical value in real-world cybersecurity applications.

cs.CR↗