arXiv subjects

Peng Wang

Publications and source records attributed to Peng Wang.

At least 253 records · Page 14Linked to original sources

arXiv2025

ScaleADFG: Affordance-based Dexterous Functional Grasping via Scalable Dataset

Dexterous functional tool-use grasping is essential for effective robotic manipulation of tools. However, existing approaches face significant challenges in efficiently constructing large-scale datasets and ensuring generalizability to everyday object scales. These issues primarily arise from size mismatches between robotic and human hands, and the diversity in real-world object scales. To address these limitations, we propose the ScaleADFG framework, which consists of a fully automated dataset construction pipeline and a lightweight grasp generation network. Our dataset introduce an affordance-based algorithm to synthesize diverse tool-use grasp configurations without expert demonstrations, allowing flexible object-hand size ratios and enabling large robotic hands (compared to human hands) to grasp everyday objects effectively. Additionally, we leverage pre-trained models to generate extensive 3D assets and facilitate efficient retrieval of object affordances. Our dataset comprising five object categories, each containing over 1,000 unique shapes with 15 scale variations. After filtering, the dataset includes over 60,000 grasps for each 2 dexterous robotic hands. On top of this dataset, we train a lightweight, single-stage grasp generation network with a notably simple loss design, eliminating the need for post-refinement. This demonstrates the critical importance of large-scale datasets and multi-scale object variant for effective training. Extensive experiments in simulation and on real robot confirm that the ScaleADFG framework exhibits strong adaptability to objects of varying scales, enhancing functional grasp stability, diversity, and generalizability. Moreover, our network exhibits effective zero-shot transfer to real-world objects. Project page is available at https://sizhe-wang.github.io/ScaleADFG_webpage

cs.RO↗

arXiv2025

High-Fidelity Single-Shot Readout and Selective Nuclear Spin Control for a Spin-1/2 Quantum Register in Diamond

Quantum networks offer a way to overcome the size and complexity limitations of single quantum devices by linking multiple nodes into a scalable architecture. Group-IV color centers in diamond, paired with long-lived nuclear spins, have emerged as promising building blocks demonstrating proof-of-concept experiments such as blind quantum computing and quantum-enhanced sensing. However, realizing a large-scale electro-nuclear register remains a major challenge. Here we establish the germanium-vacancy (GeV) center as a viable platform for such network nodes. Using correlation spectroscopy, we identify single nuclear spins within a convoluted spin environment, overcoming limitations imposed by the color center's spin-$1/2$ nature and thereby enabling indirect control of these nuclear spins. We further demonstrate high-fidelity single-shot readout of both the GeV center ($95.8\,\%$) and a neighboring ${}^{13}\text{C}$ nuclear spin ($93.7\,\%$), a key tool for feed-forward error correction. These critical advances position the GeV center as a compelling candidate for next-generation quantum network nodes.

quant-ph↗

arXiv2025

Separate the Wheat from the Chaff: Winnowing Down Divergent Views in Retrieval Augmented Generation

Retrieval-augmented generation (RAG) enhances large language models (LLMs) by integrating external knowledge sources to address their limitations in accessing up-to-date or specialized information. A natural strategy to increase the likelihood of retrieving relevant information is to expand the number of retrieved documents. However, involving more documents could introduce significant noise, as many documents may be irrelevant or misleading, thereby reducing the overall accuracy of the generated responses. To overcome the challenge associated with handling a larger number of documents, we propose WinnowRAG, a novel RAG framework designed to systematically filter out noisy documents while preserving valuable content -- a process we refer to as winnowing. WinnowRAG operates in two stages: In Stage I, we perform query-aware clustering to group similar documents and form distinct topic clusters. Each cluster is assigned to an LLM agent for generating a unique answer. In Stage II, we perform winnowing, wherein a critic LLM evaluates the outputs of multiple agents and iteratively separates useful documents from noisy ones. To retain useful documents when discarding agents, we propose two strategic merging techniques to ensure that only relevant knowledge is used for generating the final response. Crucially, WinnowRAG is model-agnostic and does not require any model fine-tuning, making it easily adaptable to various tasks. Extensive experiments on various realistic datasets demonstrate the effectiveness of WinnowRAG over state-of-the-art baselines.

cs.CL↗

arXiv2025

On-device Computation of Single-lead ECG Parameters for Real-time Remote Cardiac Health Assessment: A Real-world Validation Study

Accurate, continuous out-of-hospital electrocardiogram (ECG) parameter measurement is vital for real-time cardiac health monitoring and telemedicine. On-device computation of single-lead ECG parameters enables timely assessment without reliance on centralized data processing, advancing personalized, ubiquitous cardiac care-yet comprehensive validation across heterogeneous real-world populations remains limited. This study validated the on-device algorithm FeatureDB (https://github.com/PKUDigitalHealth/FeatureDB) using two datasets: HeartVoice-ECG-lite (369 participants with single-lead ECGs annotated by two physicians) and PTB-XL/PTB-XL+ (21,354 patients with 12-lead ECGs and physicians' diagnostic annotations). FeatureDB computed PR, QT, and QTc intervals, with accuracy evaluated against physician annotations via mean absolute error (MAE), correlation analysis, and Bland-Altman analysis. Diagnostic performance for first-degree atrioventricular block (AVBI, PR-based) and long QT syndrome (LQT, QTc-based) was benchmarked against commercial 12-lead systems (12SL, Uni-G) and open-source algorithm Deli, using AUC, accuracy, sensitivity, and specificity. Results showed high concordance with expert annotations (Pearson correlations: 0.836-0.960), MAEs matching inter-observer variability, and minimal bias. AVBI AUC reached 0.787 (12SL: 0.859; Uni-G: 0.812; Deli: 0.501); LQT AUC was 0.684 (12SL: 0.716; Uni-G: 0.605; Deli: 0.569)-comparable to commercial tools and superior to open-source alternatives. FeatureDB delivers physician-level parameter accuracy and commercial-grade abnormality detection via single-lead devices, supporting scalable telemedicine, decentralized cardiac screening, and continuous monitoring in community and outpatient settings.

eess.SP↗

arXiv2025

LyapLock: Bounded Knowledge Preservation in Sequential Large Language Model Editing

Large Language Models often contain factually incorrect or outdated knowledge, giving rise to model editing methods for precise knowledge updates. However, current mainstream locate-then-edit approaches exhibit a progressive performance decline during sequential editing, due to inadequate mechanisms for long-term knowledge preservation. To tackle this, we model the sequential editing as a constrained stochastic programming. Given the challenges posed by the cumulative preservation error constraint and the gradually revealed editing tasks, \textbf{LyapLock} is proposed. It integrates queuing theory and Lyapunov optimization to decompose the long-term constrained programming into tractable stepwise subproblems for efficient solving. This is the first model editing framework with rigorous theoretical guarantees, achieving asymptotic optimal editing performance while meeting the constraints of long-term knowledge preservation. Experimental results show that our framework scales sequential editing capacity to over 10,000 edits while stabilizing general capabilities and boosting average editing efficacy by 11.89\% over SOTA baselines. Furthermore, it can be leveraged to enhance the performance of baseline methods. Our code is released on https://github.com/caskcsg/LyapLock.

cs.CL↗

arXiv2025

Beyond the Local Group -- I: Cosmic Filaments Govern the Spatial Alignments of Galaxy Pairs

Using a large observational sample from the Sloan Digital Sky Survey, we investigate the spatial alignment between galaxy pairs and their local cosmic filaments. Focusing on pairs with stellar masses and separations comparable to the Milky Way-Andromeda (MW-M31) system, we measure the angle between the pair connecting line and the orientation of the host filament, determined using a filament catalog constructed via the Bisous model. Our analysis reveals a statistically significant tendency of galaxy pairs to align their connecting lines along their host filaments, manifesting as an overall $\sim$7\% excess of alignment angles smaller than the MW-M31 case compared to a random distribution. Crucially, the strength of this alignment exhibits a strong dependence on the distance to the filament spine. Pairs located within 0.2 Mpc from the filament spine show the strongest alignment, while those beyond 1 Mpc display no significant alignment. Furthermore, we identify a bimodal distribution of alignment angles near filament cores, suggesting distinct dynamical populations potentially associated with infall and interaction processes. Our results provide robust observational support for theoretical models in which anisotropic accretion and tidal forces within the cosmic web drive galaxy pair evolution. They also position the MW-M31 system as a representative filament-aligned pair, offering insights into Local Group assembly. This study demonstrates the cosmic web's critical role in dictating pair orientations and motivates future work with kinematic data to unravel galaxy-environment interplay.

astro-ph.GA↗

arXiv2025

Deep Learning in Dental Image Analysis: A Systematic Review of Datasets, Methodologies, and Emerging Challenges

Efficient analysis and processing of dental images are crucial for dentists to achieve accurate diagnosis and optimal treatment planning. However, dental imaging inherently poses several challenges, such as low contrast, metallic artifacts, and variations in projection angles. Combined with the subjectivity arising from differences in clinicians' expertise, manual interpretation often proves time-consuming and prone to inconsistency. Artificial intelligence (AI)-based automated dental image analysis (DIA) offers a promising solution to these issues and has become an integral part of computer-aided dental diagnosis and treatment. Among various AI technologies, deep learning (DL) stands out as the most widely applied and influential approach due to its superior feature extraction and representation capabilities. To comprehensively summarize recent progress in this field, we focus on the two fundamental aspects of DL research-datasets and models. In this paper, we systematically review 260 studies on DL applications in DIA, including 49 papers on publicly available dental datasets and 211 papers on DL-based algorithms. We first introduce the basic concepts of dental imaging and summarize the characteristics and acquisition methods of existing datasets. Then, we present the foundational techniques of DL and categorize relevant models and algorithms according to different DIA tasks, analyzing their network architectures, optimization strategies, training methods, and performance. Furthermore, we summarize commonly used training and evaluation metrics in the DIA domain. Finally, we discuss the current challenges of existing research and outline potential future directions. We hope that this work provides a valuable and systematic reference for researchers in this field. All supplementary materials and detailed comparison tables will be made publicly available on GitHub.

cs.CV↗

arXiv2025

Styl3R: Instant 3D Stylized Reconstruction for Arbitrary Scenes and Styles

Stylizing 3D scenes instantly while maintaining multi-view consistency and faithfully resembling a style image remains a significant challenge. Current state-of-the-art 3D stylization methods typically involve computationally intensive test-time optimization to transfer artistic features into a pretrained 3D representation, often requiring dense posed input images. In contrast, leveraging recent advances in feed-forward reconstruction models, we demonstrate a novel approach to achieve direct 3D stylization in less than a second using unposed sparse-view scene images and an arbitrary style image. To address the inherent decoupling between reconstruction and stylization, we introduce a branched architecture that separates structure modeling and appearance shading, effectively preventing stylistic transfer from distorting the underlying 3D scene structure. Furthermore, we adapt an identity loss to facilitate pre-training our stylization model through the novel view synthesis task. This strategy also allows our model to retain its original reconstruction capabilities while being fine-tuned for stylization. Comprehensive evaluations, using both in-domain and out-of-domain datasets, demonstrate that our approach produces high-quality stylized 3D content that achieve a superior blend of style and scene appearance, while also outperforming existing methods in terms of multi-view consistency and efficiency.

cs.CV↗

arXiv2025

Inverse designed Hamiltonians for perfect state transfer and remote entanglement generation, and applications in superconducting qubits

Hamiltonian inverse engineering enables the design of protocols for specific quantum evolutions or target state preparation. Perfect state transfer (PST) and remote entanglement generation are notable examples, as they serve as key primitives in quantum information processing. However, Hamiltonians obtained through conventional methods often lack robustness against noise. Assisted by inverse engineering, we begin with a noise-resilient energy spectrum and construct a class of Hamiltonians, referred to as the dome model, that significantly improves the system's robustness against noise, as confirmed by numerical simulations. This model introduces a tunable parameter $m$ that modifies the energy-level spacing and gives rise to a well-structured Hamiltonian. It reduces to the conventional PST model at $m=0$ and simplifies to a SWAP model involving only two end qubits in the large-$m$ regime. To address the challenge of scalability, we propose a cascaded strategy that divides long-distance PST into multiple consecutive PST steps. Our work is particularly suited for demonstration on superconducting qubits with tunable couplers, which enable rapid and flexible Hamiltonian engineering, thereby advancing the experimental potential of robust and scalable quantum information processing.

quant-ph↗

arXiv2025

How Einstein's Equation Emerges From CFT$_2$

The {\it finiteness} of the entanglement entropies between disjoint subsystems enables us to show that, the dynamical equation of the entanglement entropy in CFT$_2$ is precisely three dimensional Einstein's equation. We establish a profound relation between the cosmological constant and CFT$_2$ entanglement entropy. Thus entanglement entropies induce internal gravitational geometries in CFT$_2$. Extracting the dual metric from an entanglement entropy becomes a straightforward procedure. Remarkably, we discover that the renormalization group equation is a geometric identity.

hep-th↗

arXiv2025

ToolExpander: Extending the Frontiers of Tool-Using Reinforcement Learning to Weak LLMs

Training Large Language Models (LLMs) with Group Relative Policy Optimization (GRPO) encounters a significant challenge: models often fail to produce accurate responses, particularly in small-scale architectures. This limitation not only diminishes performance improvements and undermines the potential of GRPO but also frequently leads to mid-training collapse, adversely affecting stability and final efficacy. To address these issues, we propose ToolExpander, a novel framework that advances tool-oriented reinforcement learning for resource-constrained LLMs through two key innovations:(1) Dynamic Multi-Round Hard Sampling, which dynamically substitutes challenging samples(those without correct outputs over 10 rollouts) with high-quality few-shot demonstrations during training, coupled with an exponential learning rate decay strategy to mitigate oscillations;(2) Self-Exemplifying Thinking, an enhanced GRPO framework that eliminates KL divergence and incorporates adjusted clipping coefficients, encouraging models to autonomously generate and analyze few-shot examples via a minimal additional reward (0.01).Experimental results demonstrate that ToolExpander significantly enhances tool-using capabilities in LLMs, especially in weaker small-scale models, improving both training stability and overall performance.

cs.CL↗

arXiv2025

Sustainable Self-evolution Adversarial Training

With the wide application of deep neural network models in various computer vision tasks, there has been a proliferation of adversarial example generation strategies aimed at deeply exploring model security. However, existing adversarial training defense models, which rely on single or limited types of attacks under a one-time learning process, struggle to adapt to the dynamic and evolving nature of attack methods. Therefore, to achieve defense performance improvements for models in long-term applications, we propose a novel Sustainable Self-Evolution Adversarial Training (SSEAT) framework. Specifically, we introduce a continual adversarial defense pipeline to realize learning from various kinds of adversarial examples across multiple stages. Additionally, to address the issue of model catastrophic forgetting caused by continual learning from ongoing novel attacks, we propose an adversarial data replay module to better select more diverse and key relearning data. Furthermore, we design a consistency regularization strategy to encourage current defense models to learn more from previously trained ones, guiding them to retain more past knowledge and maintain accuracy on clean samples. Extensive experiments have been conducted to verify the efficacy of the proposed SSEAT defense method, which demonstrates superior defense performance and classification accuracy compared to competitors.Code is available at https://github.com/aup520/SSEAT

cs.CV↗

arXiv2025

The Safety Challenge of World Models for Embodied AI Agents: A Review

The rapid progress in embodied artificial intelligence has highlighted the necessity for more advanced and integrated models that can perceive, interpret, and predict environmental dynamics. In this context, World Models (WMs) have been introduced to provide embodied agents with the abilities to anticipate future environmental states and fill in knowledge gaps, thereby enhancing agents' ability to plan and execute actions. However, when dealing with embodied agents it is fundamental to ensure that predictions are safe for both the agent and the environment. In this article, we conduct a comprehensive literature review of World Models in the domains of autonomous driving and robotics, with a specific focus on the safety implications of scene and control generation tasks. Our review is complemented by an empirical analysis, wherein we collect and examine predictions from state-of-the-art models, identify and categorize common faults (herein referred to as pathologies), and provide a quantitative evaluation of the results.

cs.AI↗

arXiv2025

High temperature Neel skyrmions in simple ferromagnets

A wide variety of chiral non-collinear spin textures have been discovered and have unique properties that make them highly interesting for technological applications. However, many of these are found in complex materials and only in a narrow window of temperature. Here, we show the formation of Neel-type skyrmions in thin layers of simple ferromagnetic alloys, namely Co-Al and Co-Ni-Al, over a wide range of temperature up to 770 K, by imposing a vertical strain gradient via epitaxy with an Ir-Al underlayer. The Neel skyrmions are directly observed using Lorentz transmission electron microscopy in freestanding membranes at high temperatures and the strain gradient is directly measured from x-ray diffraction anomalous peak profiles. Our concept allows simple centrosymmetric ferromagnets with high magnetic ordering temperatures to exhibit hot skyrmions, thereby, bringing closer skyrmionic electronics.

cond-mat.mtrl-sci↗

arXiv2025

MEMTRACK: Evaluating Long-Term Memory and State Tracking in Multi-Platform Dynamic Agent Environments

Recent works on context and memory benchmarking have primarily focused on conversational instances but the need for evaluating memory in dynamic enterprise environments is crucial for its effective application. We introduce MEMTRACK, a benchmark designed to evaluate long-term memory and state tracking in multi-platform agent environments. MEMTRACK models realistic organizational workflows by integrating asynchronous events across multiple communication and productivity platforms such as Slack, Linear and Git. Each benchmark instance provides a chronologically platform-interleaved timeline, with noisy, conflicting, cross-referring information as well as potential codebase/file-system comprehension and exploration. Consequently, our benchmark tests memory capabilities such as acquistion, selection and conflict resolution. We curate the MEMTRACK dataset through both manual expert driven design and scalable agent based synthesis, generating ecologically valid scenarios grounded in real world software development processes. We introduce pertinent metrics for Correctness, Efficiency, and Redundancy that capture the effectiveness of memory mechanisms beyond simple QA performance. Experiments across SoTA LLMs and memory backends reveal challenges in utilizing memory across long horizons, handling cross-platform dependencies, and resolving contradictions. Notably, the best performing GPT-5 model only achieves a 60\% Correctness score on MEMTRACK. This work provides an extensible framework for advancing evaluation research for memory-augmented agents, beyond existing focus on conversational setups, and sets the stage for multi-agent, multi-platform memory benchmarking in complex organizational settings

cs.AI↗

arXiv2025

Repro Samples Method for Model-Free Inference in High-Dimensional Binary Classification

This paper presents a novel method for statistical inference in high-dimensional binary models with unspecified structure, where we leverage a (potentially misspecified) sparsity-constrained working generalized linear model (GLM) to facilitate the inference process. Our method is based on the repro samples framework, which generates artificial samples that mimic the actual data-generating process. Our inference targets include the model support, case probabilities, and the oracle regression coefficients defined in the working GLM. The proposed method has three major advantages. First, this approach is model-free, that is, it does not rely on specific model assumptions such as logistic or probit regression, nor does it require sparsity assumptions on the underlying model. Second, for model support, we construct a model candidate set for the most influential covariates that achieves guaranteed coverage under a weak signal strength assumption. Third, for oracle regression coefficients, we establish confidence sets for any group of linear combinations of regression coefficients. Simulation results demonstrate that the proposed method produces valid and small model candidate sets. It also achieves better coverage for regression coefficients than the state-of-the-art debiasing methods when the working model is the actual model that generates the sample data. Additionally, we analyze single-cell RNA-seq data on the immune response. Besides identifying genes previously proven as relevant in the literature, our method also discovers a significant gene that has not been studied before, revealing a potential new direction in understanding cellular immune response mechanisms.

stat.ME↗

arXiv2025

Observational Evidence for Spin Alignment Between Galaxy Groups and Their Central Galaxies

We present a systematic observational study of the spin alignment between galaxy groups and their central galaxies using a large spectroscopic sample from the Sloan Digital Sky Survey. Unlike previous studies based on morphology or major axis alignment, we use spin as a direct, dynamically motivated probe. To match the limitations of observational data, the group spin, central-galaxy spin, and alignment angle are defined and measured in projection on the sky plane. By applying two novel spin estimators for galaxy groups, we find strong statistical evidence for a preferential alignment, with a mean projected angle of $34.17^\circ \pm 0.29^\circ$, significantly deviating from random expectation at $37.77σ$. This alignment signal persists across a wide range of group and galaxy properties, but its strength is modulated by mass, morphology, and color. Specifically, we find that more massive groups and more massive central galaxies exhibit stronger alignment. Furthermore, elliptical central galaxies show stronger alignment than spirals, and bluer central galaxies are more strongly aligned than redder ones. Our results suggest a close dynamical link between the spin of central galaxies and their host groups, modulated by their physical properties and star formation history. These results provide new insights into the dynamical connection between central galaxies and their host dark matter halos.

astro-ph.GA↗

arXiv2025

RapidMV: Leveraging Spatio-Angular Representations for Efficient and Consistent Text-to-Multi-View Synthesis

Generating synthetic multi-view images from a text prompt is an essential bridge to generating synthetic 3D assets. In this work, we introduce RapidMV, a novel text-to-multi-view generative model that can produce 32 multi-view synthetic images in just around 5 seconds. In essence, we propose a novel spatio-angular latent space, encoding both the spatial appearance and angular viewpoint deviations into a single latent for improved efficiency and multi-view consistency. We achieve effective training of RapidMV by strategically decomposing our training process into multiple steps. We demonstrate that RapidMV outperforms existing methods in terms of consistency and latency, with competitive quality and text-image alignment.

cs.CV↗