arXiv subjects

Jun Yan

Publications and source records attributed to Jun Yan.

At least 91 records · Page 5Linked to original sources

arXiv2025

Enhancing LiDAR Point Features with Foundation Model Priors for 3D Object Detection

Recent advances in foundation models have opened up new possibilities for enhancing 3D perception. In particular, DepthAnything offers dense and reliable geometric priors from monocular RGB images, which can complement sparse LiDAR data in autonomous driving scenarios. However, such priors remain underutilized in LiDAR-based 3D object detection. In this paper, we address the limited expressiveness of raw LiDAR point features, especially the weak discriminative capability of the reflectance attribute, by introducing depth priors predicted by DepthAnything. These priors are fused with the original LiDAR attributes to enrich each point's representation. To leverage the enhanced point features, we propose a point-wise feature extraction module. Then, a Dual-Path RoI feature extraction framework is employed, comprising a voxel-based branch for global semantic context and a point-based branch for fine-grained structural details. To effectively integrate the complementary RoI features, we introduce a bidirectional gated RoI feature fusion module that balances global and local cues. Extensive experiments on the KITTI benchmark show that our method consistently improves detection accuracy, demonstrating the value of incorporating visual foundation model priors into LiDAR-based 3D object detection.

cs.CV↗

arXiv2025

Write Your Own CodeChecker: An Automated Test-Driven Checker Development Approach with LLMs

With the rising demand for code quality assurance, developers are not only utilizing existing static code checkers but also seeking custom checkers to satisfy their specific needs. Nowadays, various code-checking frameworks provide extensive checker customization interfaces to meet this need. However, both the abstract checking logic and the complex API usage of large-scale checker frameworks make this task challenging. To this end, automated code checker generation is anticipated to ease the burden of checker development. In this paper, we propose AutoChecker, an innovative LLM-powered approach that can write code checkers automatically based on only a rule description and a test suite. To achieve comprehensive checking logic, AutoChecker incrementally updates the checker's logic by focusing on solving one selected case each time. To obtain precise API knowledge, during each iteration, it leverages fine-grained logic-guided API-context retrieval, where it first decomposes the checking logic into a series of sub-operations and then retrieves checker-related API-contexts for each sub-operation. For evaluation, we apply AutoChecker, five baselines, and three ablation methods using multiple LLMs to generate checkers for 20 randomly selected PMD rules. Experimental results show that AutoChecker significantly outperforms others across all effectiveness metrics, with an average test pass rate of 82.28%. Additionally, the checkers generated by AutoChecker can be successfully applied to real-world projects, matching the performance of official checkers.

cs.SE↗

arXiv2025

Position: Emergent Machina Sapiens Urge Rethinking Multi-Agent Paradigms

Artificial Intelligence (AI) agents capable of autonomous learning and independent decision-making hold great promise for addressing complex challenges across various critical infrastructure domains, including transportation, energy systems, and manufacturing. However, the surge in the design and deployment of AI systems, driven by various stakeholders with distinct and unaligned objectives, introduces a crucial challenge: How can uncoordinated AI systems coexist and evolve harmoniously in shared environments without creating chaos or compromising safety? To address this, we advocate for a fundamental rethinking of existing multi-agent frameworks, such as multi-agent systems and game theory, which are largely limited to predefined rules and static objective structures. We posit that AI agents should be empowered to adjust their objectives dynamically, make compromises, form coalitions, and safely compete or cooperate through evolving relationships and social feedback. Through two case studies in critical infrastructure applications, we call for a shift toward the emergent, self-organizing, and context-aware nature of these multi-agentic AI systems.

cs.MA↗

arXiv2025

FedSC: Federated Learning with Semantic-Aware Collaboration

Federated learning (FL) aims to train models collaboratively across clients without sharing data for privacy-preserving. However, one major challenge is the data heterogeneity issue, which refers to the biased labeling preferences at multiple clients. A number of existing FL methods attempt to tackle data heterogeneity locally (e.g., regularizing local models) or globally (e.g., fine-tuning global model), often neglecting inherent semantic information contained in each client. To explore the possibility of using intra-client semantically meaningful knowledge in handling data heterogeneity, in this paper, we propose Federated Learning with Semantic-Aware Collaboration (FedSC) to capture client-specific and class-relevant knowledge across heterogeneous clients. The core idea of FedSC is to construct relational prototypes and consistent prototypes at semantic-level, aiming to provide fruitful class underlying knowledge and stable convergence signals in a prototype-wise collaborative way. On the one hand, FedSC introduces an inter-contrastive learning strategy to bring instance-level embeddings closer to relational prototypes with the same semantics and away from distinct classes. On the other hand, FedSC devises consistent prototypes via a discrepancy aggregation manner, as a regularization penalty to constrain the optimization region of the local model. Moreover, a theoretical analysis for FedSC is provided to ensure a convergence guarantee. Experimental results on various challenging scenarios demonstrate the effectiveness of FedSC and the efficiency of crucial components.

cs.CV↗

arXiv2025

RefPentester: A Knowledge-Informed Self-Reflective Penetration Testing Framework Based on Large Language Models

Automated penetration testing (AutoPT) powered by large language models (LLMs) has gained attention for its ability to automate ethical hacking processes and identify vulnerabilities in target systems by leveraging the inherent knowledge of LLMs. However, existing LLM-based AutoPT frameworks often underperform compared to human experts in challenging tasks for several reasons: the imbalanced knowledge used in LLM training, short-sightedness in the planning process, and hallucinations during command generation. Moreover, the trial-and-error nature of the PT process is constrained by existing frameworks lacking mechanisms to learn from previous failures, restricting adaptive improvement of PT strategies. To address these limitations, we propose a knowledge-informed, self-reflective PT framework powered by LLMs, called RefPentester. This AutoPT framework is designed to assist human operators in identifying the current stage of the PT process, selecting appropriate tactics and techniques for each stage, choosing suggested actions, providing step-by-step operational guidance, and reflecting on and learning from previous failed operations. We also modeled the PT process as a seven-state Stage Machine to integrate the proposed framework effectively. The evaluation shows that RefPentester can successfully reveal credentials on Hack The Box's Sau machine, outperforming the baseline GPT-4o model by 16.7%. Across PT stages, RefPentester also demonstrates superior success rates on PT stage transitions.

cs.AI↗

arXiv2025

On Devon Allen's Disqualification at the 2022 World Track and Field Championships

Devon Allen's disqualification at the men's 110-meter hurdle final at the 2022 World Track and Field Championships, due to a reaction time (RT) of 0.099 seconds-just 0.001 seconds below the allowable threshold-sparked widespread debate over the fairness and validity of RT rules. This study investigates two key issues: variations in timing systems and the justification for the 0.1-second disqualification threshold. We pooled RT data from men's 110-meter hurdles and 100-meter dash, as well as women's 100-meter hurdles and 100-meter dash, spanning national and international competitions. Using a rank-sum test for clustered data, we compared RTs across multiple competitions, while a generalized Gamma model with random effects for venue and heat was applied to evaluate the threshold. Our analyses reveal significant differences in RTs between the 2022 World Championships and other competitions, pointing to systematic variations in timing systems. Additionally, the model shows that RTs be low 0.1 seconds, though rare, are physiologically plausible. These findings highlight the need for standardized timing protocols and a re-evaluation of the 0.1-second disqualification threshold to promote fairness in elite competition.

stat.AP↗

arXiv2025

HACMony: Automatically Detecting Hopping-related Audio-stream Conflict Issues on HarmonyOS

HarmonyOS is emerging as a popular distributed operating system for diverse mobile devices. One of its standout features is app-hopping, which allows users to seamlessly transition apps across different HarmonyOS devices. However, when apps playing audio streams hop between devices, they can easily trigger Hopping-related Audio-stream Conflict (HAC) scenarios. Improper resolution of HAC will lead to significant HAC issues, which are harder to detect compared to single-device audio-stream conflicts, due to the unclear semantics of HarmonyOS's app-hopping mechanism and the lack of effective multi-app hopping testing methods. To fill the gap, this paper introduces an automated and efficient approach to detecting HAC issues. We formalized the operational semantics of HarmonyOS's app-hopping mechanism for audio streams for the first time. Leveraging this formalization, we designed an Audio Service Transition Graph (ASTG) to model the behaviors of audio-API-related services and proposed a model-based approach to detect HAC issues automatically. Our techniques were implemented in a tool, HACMony, and evaluated on 20 real-world HarmonyOS apps. Experimental results reveal that 11 of the 20 apps exhibit HAC issues. Additionally, we summarized the detected issues into two typical types, namely MoD and MoR, and analyzed their characteristics to assist and guide both app and OS developers.

cs.SE↗

arXiv2025

FedSKC: Federated Learning with Non-IID Data via Structural Knowledge Collaboration

With the advancement of edge computing, federated learning (FL) displays a bright promise as a privacy-preserving collaborative learning paradigm. However, one major challenge for FL is the data heterogeneity issue, which refers to the biased labeling preferences among multiple clients, negatively impacting convergence and model performance. Most previous FL methods attempt to tackle the data heterogeneity issue locally or globally, neglecting underlying class-wise structure information contained in each client. In this paper, we first study how data heterogeneity affects the divergence of the model and decompose it into local, global, and sampling drift sub-problems. To explore the potential of using intra-client class-wise structural knowledge in handling these drifts, we thus propose Federated Learning with Structural Knowledge Collaboration (FedSKC). The key idea of FedSKC is to extract and transfer domain preferences from inter-client data distributions, offering diverse class-relevant knowledge and a fair convergent signal. FedSKC comprises three components: i) local contrastive learning, to prevent weight divergence resulting from local training; ii) global discrepancy aggregation, which addresses the parameter deviation between the server and clients; iii) global period review, correcting for the sampling drift introduced by the server randomly selecting devices. We have theoretically analyzed FedSKC under non-convex objectives and empirically validated its superiority through extensive experimental results.

cs.LG↗

arXiv2025

Ionization potential depression model with the influence of neighboring ions for warm/hot and dense plasma

For warm or hot and dense plasma, ionization potential depression plays a crucial role in determining the ionization balance and understanding the resulting microscopic plasma properties. However, the applicability of the widely used IPD models is currently limited under WDP conditions, where the influence of neighboring ions on IPD becomes nonnegligible. Neighboring ions can directly influence the screening potential around the target ion, which then changes the ionization potential. Furthermore, similar to solid-state systems, outer atomic orbitals expand into continuous energy bands due to the existence of neighboring ions, and electrons in these continuous bands can travel from target ion into neighboring ions and become delocalized. As a result, even for their total energy E<0, electrons excited into these continuous bands can be considered ionized, and the ionization conditions differ from those in isolated situations. In our previous work with an atomic state dependent screening model, we included the influence of temporarily recombined electron distributions due to inelastic collision processes between plasma electrons and ions, and evident contributions from these electrons to the screening potential were found under WDP conditions. We now further incorporate the direct contributions of neighboring ions to both screening potentials and ionization conditions. This extended framework reveals that the contribution from neighboring ions substantially influences IPD in WDP. The developed model demonstrates good agreement with experiments for Al, Mg, and Si plasmas with a wide range of 70 to 700 eV temperatures and 1 to 3 times the solid density as well as the hollow Al ions measured in the experiment.

physics.plasm-ph↗

arXiv2025

From LLMs to LLM-based Agents for Software Engineering: A Survey of Current, Challenges and Future

With the rise of large language models (LLMs), researchers are increasingly exploring their applications in var ious vertical domains, such as software engineering. LLMs have achieved remarkable success in areas including code generation and vulnerability detection. However, they also exhibit numerous limitations and shortcomings. LLM-based agents, a novel tech nology with the potential for Artificial General Intelligence (AGI), combine LLMs as the core for decision-making and action-taking, addressing some of the inherent limitations of LLMs such as lack of autonomy and self-improvement. Despite numerous studies and surveys exploring the possibility of using LLMs in software engineering, it lacks a clear distinction between LLMs and LLM based agents. It is still in its early stage for a unified standard and benchmarking to qualify an LLM solution as an LLM-based agent in its domain. In this survey, we broadly investigate the current practice and solutions for LLMs and LLM-based agents for software engineering. In particular we summarise six key topics: requirement engineering, code generation, autonomous decision-making, software design, test generation, and software maintenance. We review and differentiate the work of LLMs and LLM-based agents from these six topics, examining their differences and similarities in tasks, benchmarks, and evaluation metrics. Finally, we discuss the models and benchmarks used, providing a comprehensive analysis of their applications and effectiveness in software engineering. We anticipate this work will shed some lights on pushing the boundaries of LLM-based agents in software engineering for future research.

cs.SE↗

arXiv2025

Adversarial Examples in Environment Perception for Automated Driving (Review)

The renaissance of deep learning has led to the massive development of automated driving. However, deep neural networks are vulnerable to adversarial examples. The perturbations of adversarial examples are imperceptible to human eyes but can lead to the false predictions of neural networks. It poses a huge risk to artificial intelligence (AI) applications for automated driving. This survey systematically reviews the development of adversarial robustness research over the past decade, including the attack and defense methods and their applications in automated driving. The growth of automated driving pushes forward the realization of trustworthy AI applications. This review lists significant references in the research history of adversarial examples.

cs.CV↗

arXiv2025

WS-DETR: Robust Water Surface Object Detection through Vision-Radar Fusion with Detection Transformer

Robust object detection for Unmanned Surface Vehicles (USVs) in complex water environments is essential for reliable navigation and operation. Specifically, water surface object detection faces challenges from blurred edges and diverse object scales. Although vision-radar fusion offers a feasible solution, existing approaches suffer from cross-modal feature conflicts, which negatively affect model robustness. To address this problem, we propose a robust vision-radar fusion model WS-DETR. In particular, we first introduce a Multi-Scale Edge Information Integration (MSEII) module to enhance edge perception and a Hierarchical Feature Aggregator (HiFA) to boost multi-scale object detection in the encoder. Then, we adopt self-moving point representations for continuous convolution and residual connection to efficiently extract irregular features under the scenarios of irregular point cloud data. To further mitigate cross-modal conflicts, an Adaptive Feature Interactive Fusion (AFIF) module is introduced to integrate visual and radar features through geometric alignment and semantic fusion. Extensive experiments on the WaterScenes dataset demonstrate that WS-DETR achieves state-of-the-art (SOTA) performance, maintaining its superiority even under adverse weather and lighting conditions.

cs.CV↗

arXiv2025

A Universal Model Combining Differential Equations and Neural Networks for Ball Trajectory Prediction

This paper presents a data driven universal ball trajectory prediction method integrated with physics equations. Existing methods are designed for specific ball types and struggle to generalize. This challenge arises from three key factors. First, learning-based models require large datasets but suffer from accuracy drops in unseen scenarios. Second, physics-based models rely on complex formulas and detailed inputs, yet accurately obtaining ball states, such as spin, is often impractical. Third, integrating physical principles with neural networks to achieve high accuracy, fast inference, and strong generalization remains difficult. To address these issues, we propose an innovative approach that incorporates physics-based equations and neural networks. We first derive three generalized physical formulas. Then, using a neural network and observed trajectory points, we infer certain parameters while fitting the remaining ones. These formulas enable precise trajectory prediction with minimal training data: only a few dozen samples. Extensive experiments demonstrate our method superiority in generalization, real-time performance, and accuracy.

cs.LG↗

arXiv2025

An improved hypergraph Mantel's Theorem

In a recent paper, Chao and Yu used an entropy method to show that the Turán density of a certain family $\mathcal{F}$ of $\lfloor r/2\rfloor$ triangle-like $r$-uniform hypergraphs is $r!/r^r$. Later, Liu determined for large $n$ the exact Turán number $\text{ex}(n,\mathcal{F})$ of this family, and showed that the unique extremal graph is the balanced complete $r$-partite $r$-uniform hypergraph. These two results together can be viewed as a hypergraph version of Mantel's Theorem. In this paper, building on their methods, we improve both of these results by showing that they still hold with a subfamily $\mathcal{F}'\subset\mathcal{F}$ of size $\lceil r/e\rceil$ in place of $\mathcal{F}$.

math.CO↗

arXiv2025

Magnet: Multi-turn Tool-use Data Synthesis and Distillation via Graph Translation

Large language models (LLMs) have exhibited the ability to effectively utilize external tools to address user queries. However, their performance may be limited in complex, multi-turn interactions involving users and multiple tools. To address this, we propose Magnet, a principled framework for synthesizing high-quality training trajectories to enhance the function calling capability of large language model agents in multi-turn conversations with humans. The framework is based on automatic and iterative translations from a function signature path to a sequence of queries and executable function calls. We model the complicated function interactions in multi-turn cases with graph and design novel node operations to build reliable signature paths. Motivated by context distillation, when guiding the generation of positive and negative trajectories using a teacher model, we provide reference function call sequences as positive hints in context and contrastive, incorrect function calls as negative hints. Experiments show that training with the positive trajectories with supervised fine-tuning and preference optimization against negative trajectories, our 14B model, Magnet-14B-mDPO, obtains 68.01 on BFCL-v3 and 73.30 on ToolQuery, surpassing the performance of the teacher model Gemini-1.5-pro-002 by a large margin in function calling.

cs.CL↗

arXiv2025

Principles for Open Data Curation: A Case Study with the New York City 311 Service Request Data

In the early 21st century, the open data movement began to transform societies and governments by promoting transparency, innovation, and public engagement. The City of New York (NYC) has been at the forefront of this movement since the enactment of the Open Data Law in 2012, creating the NYC Open Data portal. The portal currently hosts 2,700 datasets, serving as a crucial resource for research across various domains, including health, urban development, and transportation. However, the effective use of open data relies heavily on data quality and usability, challenges that remain insufficiently addressed in the literature. This paper examines these challenges via a case study of the NYC 311 Service Request dataset, identifying key issues in data validity, consistency, and curation efficiency. We propose a set of data curation principles, tailored for government-released open data, to address these challenges. Our findings highlight the importance of harmonized field definitions, streamlined storage, and automated quality checks, offering practical guidelines for improving the reliability and utility of open datasets.

cs.DB↗

arXiv2025

Highly tunable valley polarization of potential-trapped moiré excitons in WSe2/WS2 heterojunctions

Moiré superlattices created by stacking atomic layers of transition metal dichalcogenide semiconductors have emerged as a class of fascinating artificial photonic and electronic materials. An appealing attribute of these structures is the inheritance of the valley degree of freedom from the constituent monolayers. Recent studies show evidence that the valley polarization of the moiré excitons is highly tunable. In heterojunctions of WSe2/WS2, marked improvement in valley polarization is observed by increasing optical excitation power, a behavior that is quite distinct from the monolayers, and lacks a clear understanding so far. In this work, we show that this highly tunable valley property arises from filling of the moiré superlattice, which provides an intriguing mechanism for engineering these quantum opto-valleytronic platforms. Our data further demonstrate that the long-range electron-hole exchange interaction, despite being significantly weakened in the junctions, is the dominant source of moiré exciton intervalley scattering at low population. Using magnetic field tuning, we quantitatively determine the exchange interaction strength to be 0.03 meV and 0.24 meV for 0- and 60-degrees twisted samples respectively in our experiments, about one order of magnitude weaker than that in the monolayers.

cond-mat.mes-hall↗

arXiv2025

Air-stable lithiation engineering of $\mathrm{MoS}_{2}$ for direct-bandgap multilayers

Due to its sizable direct bandgap and strong light-matter interactions, the preparation of monolayer $\mathrm{MoS}_{2}$ has attracted significant attention and intensive research efforts. However, multilayer $\mathrm{MoS}_{2}$ is largely overlooked because of its optically inactive indirect bandgap caused by interlayer coupling. It is highly desirable to modulate and decrease the interlayer coupling so that each layer in multilayer $\mathrm{MoS}_{2}$ can exhibit a monolayer-like direct-gap behavior. Here, we demonstrate the nanoprobe fabrication of $\mathrm{Li}_{x}\mathrm{MoS}_{2}$-based multilayers exhibiting a direct bandgap and strong photoluminescence emission from tightly bound excitons and trions. The fabrication is facilitated by our newly developed Li-ion platform, featuring tip-induced Li intercalation, air stability and rewritability. Raman characterizations reveal that controlled Li intercalation effectively transforms multilayer $\mathrm{MoS}_{2}$ into the stack of multiple monolayers, leading to a 26-fold enhancement of photoluminescence, compared to a monolayer. This intercalation result is different from existing observations of transforming $\mathrm{MoS}_{2}$ multilayers into metallic phases.

cond-mat.mes-hall↗