Searcharxiv⌕ Search

arXiv subjects

Jing Yu

Publications and source records attributed to Jing Yu.

At least 55 records · Page 3Linked to original sources
arXiv2025

Safe Control of Multi-Agent Systems with Minimal Communication

In many multi-agent systems, communication is limited by bandwidth, latency, and energy constraints. Designing controllers that achieve coordination and safety with minimal communication is critical for scalable and reliable deployment. This paper presents a method for designing controllers that minimize inter-agent communication in multi-agent systems while satisfying safety and coordination requirements, while conforming to communication delay constraints. The control synthesis problem is cast as a rank minimization problem, where a convex relaxation is obtained via system level synthesis. Simulation results on various tasks, including trajectory tracking with relative and heterogeneous sensing, demonstrate that the proposed method significantly reduces inter-agent transmission compared to baseline approaches.

eess.SY↗
arXiv2025

Embedding Borel graphs into grids of asymptotically optimal dimension

Let $G$ be a Borel graph all of whose finite subgraphs embed into the $d$-dimensional grid with diagonals. We show that then $G$ itself admits a Borel embedding into the Schreier graph of a free Borel action of $\mathbb Z^{O(d)}$. This strengthens an earlier result of the authors, in which $O(d)$ is replaced by $O(ρ\log ρ)$, where $ρ$ is the polynomial growth rate of $G$.

math.CO↗
arXiv2025

RecGPT-V2 Technical Report

Large language models (LLMs) have demonstrated remarkable potential in transforming recommender systems from implicit behavioral pattern matching to explicit intent reasoning. While RecGPT-V1 successfully pioneered this paradigm by integrating LLM-based reasoning into user interest mining and item tag prediction, it suffers from four fundamental limitations: (1) computational inefficiency and cognitive redundancy across multiple reasoning routes; (2) insufficient explanation diversity in fixed-template generation; (3) limited generalization under supervised learning paradigms; and (4) simplistic outcome-focused evaluation that fails to match human standards. To address these challenges, we present RecGPT-V2 with four key innovations. First, a Hierarchical Multi-Agent System restructures intent reasoning through coordinated collaboration, eliminating cognitive duplication while enabling diverse intent coverage. Combined with Hybrid Representation Inference that compresses user-behavior contexts, our framework reduces GPU consumption by 60% and improves exclusive recall from 9.39% to 10.99%. Second, a Meta-Prompting framework dynamically generates contextually adaptive prompts, improving explanation diversity by +7.3%. Third, constrained reinforcement learning mitigates multi-reward conflicts, achieving +24.1% improvement in tag prediction and +13.0% in explanation acceptance. Fourth, an Agent-as-a-Judge framework decomposes assessment into multi-step reasoning, improving human preference alignment. Online A/B tests on Taobao demonstrate significant improvements: +2.98% CTR, +3.71% IPV, +2.19% TV, and +11.46% NER. RecGPT-V2 establishes both the technical feasibility and commercial viability of deploying LLM-powered intent reasoning at scale, bridging the gap between cognitive exploration and industrial utility.

cs.IR↗
arXiv2025

Hopf algebras with the dual Chevalley property of finite corepresentation type

Let $H$ be a finite-dimensional Hopf algebra over an algebraically closed field $\Bbbk$ with the dual Chevalley property. We prove that $H$ is of finite corepresentation type if and only if it is coNakayama, if and only if the link quiver $\mathrm{Q}(H)$ of $H$ is a disjoint union of basic cycles, if and only if the link-indecomposable component $H_{(1)}$ containing $\Bbbk1$ is a pointed Hopf algebra and the link quiver of $H_{(1)}$ is a basic cycle.

math.QA↗
arXiv2025

Hopf algebras with the dual Chevalley property of discrete corepresentation type

We try to classify Hopf algebras with the dual Chevalley property of discrete corepresentation type over an algebraically closed field $\Bbb{k}$ with characteristic 0. For such Hopf algebra $H$, we characterize the link quiver of $H$ and determine the structures of the link-indecomposable component $H_{(1)}$ containing $\Bbb{k}1$. Besides, we construct an infinite-dimensional non-pointed non-cosemisimple link-indecomposable Hopf algebra $H(e_{\pm 1}, f_{\pm 1}, u, v)$ with the dual Chevalley property of discrete corepresentation type.

math.QA↗
arXiv2025

Coquasitriangular structures on Hopf algebras constructed via abelian extensions

The aim of this paper is to study coquasitriangular structures on a class of cosemisimple Hopf algebras of the form $\Bbbk^G {}^τ\#_σ \Bbbk F$, constructed as abelian extensions of $\Bbbk F$ by $\Bbbk^G$ for a finite group $G$ and an arbitrary group $F$. We investigate when a coquasitriangular structure exists on $\Bbbk^G{}^τ\#_σ\Bbbk F$ and provide characterizations of its coquasitriangular structures. As an application, we study the coquasitriangular structures for the case where $G = \mathbb{Z}_2$.

math.QA↗
arXiv2025

A Vision-Language Pre-training Model-Guided Approach for Mitigating Backdoor Attacks in Federated Learning

Defending backdoor attacks in Federated Learning (FL) under heterogeneous client data distributions encounters limitations balancing effectiveness and privacy-preserving, while most existing methods highly rely on the assumption of homogeneous client data distributions or the availability of a clean serve dataset. In this paper, we propose an FL backdoor defense framework, named CLIP-Fed, that utilizes the zero-shot learning capabilities of vision-language pre-training models. Our scheme overcomes the limitations of Non-IID imposed on defense effectiveness by integrating pre-aggregation and post-aggregation defense strategies. CLIP-Fed aligns the knowledge of the global model and CLIP on the augmented dataset using prototype contrastive loss and Kullback-Leibler divergence, so that class prototype deviations caused by backdoor samples are ensured and the correlation between trigger patterns and target labels is eliminated. In order to balance privacy-preserving and coverage enhancement of the dataset against diverse triggers, we further construct and augment the server dataset via using the multimodal large language model and frequency analysis without any client samples. Extensive experiments on representative datasets evidence the effectiveness of CLIP-Fed. Comparing to other existing methods, CLIP-Fed achieves an average reduction in Attack Success Rate, {\em i.e.}, 2.03\% on CIFAR-10 and 1.35\% on CIFAR-10-LT, while improving average Main Task Accuracy by 7.92\% and 0.48\%, respectively. Our codes are available at https://anonymous.4open.science/r/CLIP-Fed.

cs.LG↗
arXiv2025

Maximum in-general-position set in a random subset of $\mathbb{F}^d_q$

Let $α(\mathbb{F}_q^{d},p)$ be the maximum possible size of a point set in general position in a $p$-random subset of $\mathbb{F}_q^d$. We determine the order of magnitude of $α(\mathbb{F}_q^{d},p)$ up to a polylogarithmic factor by proving the balanced supersaturation conjecture of Balogh and Luo. Our result also resolves a conjecture implicitly posed by the first author, Liu, the second author and Zeng. In the course of our proof, we establish a lemma that demonstrates a ``structure vs. randomness'' phenomenon for point sets in finite-field linear spaces, which may be of independent interest.

math.CO↗
arXiv2025

Watermarking Visual Concepts for Diffusion Models

The personalization techniques of diffusion models succeed in generating images with specific concepts. This ability also poses great threats to copyright protection and network security since malicious users can generate unauthorized content and disinformation relevant to a target concept. Model watermarking is an effective solution to trace the malicious generated images and safeguard their copyright. However, existing model watermarking techniques merely achieve image-level tracing without concept traceability. When tracing infringing or harmful concepts, current approaches execute image concept detection and model tracing sequentially, where performance is critically constrained by concept detection accuracy. In this paper, we propose a lightweight concept watermarking framework that efficiently binds target concepts to model watermarks, supporting simultaneous concept identification and model tracing via single-stage watermark verification. To further enhance the robustness of concept watermarking, we propose an adversarial perturbation injection method collaboratively embedded with watermarks during image generation, avoiding watermark removal by model purification attacks. Experimental results demonstrate that ConceptWM significantly outperforms state-of-the-art watermarking methods, improving detection accuracy by 6.3%-19.3% across diverse datasets including COCO and StableDiffusionDB. Additionally, ConceptWM possesses a critical capability absent in other watermarking methods: it sustains a 21.7% FID/CLIP degradation under adversarial fine-tuning of Stable Diffusion models on WikiArt and CelebA-HQ, demonstrating its capability to mitigate model misuse.

cs.CR↗
arXiv2025

System-Level Performance and Communication Tradeoff in Networked Control with Predictions

Distributed control of large-scale systems is challenging due to the need for scalable and localized communication and computation. In this work, we introduce a Predictive System-Level Synthesis PredSLS framework that designs controllers by jointly integrating communication constraints and local disturbance predictions into an affine feedback structure. Rather than focusing on the worst-case uncertainty, PredSLS leverages both current state feedback and future system disturbance predictions to achieve distributed control of networked systems. In particular, PredSLS enables a unified system synthesis of the optimal $κ$-localized controller, therefore outperforms approaches with post hoc communication truncation, as was commonly seen in the literature. The PredSLS framework can be naturally decomposed into spatial and temporal components for efficient and parallelizable computation across the network, yielding a regret upper bound that explicitly depends on the prediction error and communication range. Our regret analysis not only reveals a non-monotonic trade-off between control performance and communication range when prediction errors are present, but also guides the identification of an optimal size for local communication neighborhoods, thereby enabling the co-design of controller and its underlying communication topology.

eess.SY↗
arXiv2025

Topic Over Source: The Key to Effective Data Mixing for Language Models Pre-training

The performance of large language models (LLMs) is significantly affected by the quality and composition of their pre-training data, which is inherently diverse, spanning various languages, sources, and topics. Effectively integrating these heterogeneous data groups is crucial for optimizing LLM performance. Previous research has predominantly concentrated on source-based data mixing, often neglecting the nuanced topic-level characteristics of the data. To address this gap, we propose a topic-based data mixing strategy that utilizes detailed topic labels generated through a multi-stage process combining unsupervised clustering, LLM-based summarization, and supervised classifier training. With this strategy, we conduct the first comprehensive comparison of topic-based versus source-based partitioning across multiple mixing strategies. We demonstrate that language models pretrained on data mixed by topics consistently outperform those trained on data mixed by sources across multiple methods including RegMix, DoReMi,temperature-based sampling, and a manual mixing method based on downstream task performance. Our theoretical analysis reveals that topic-based data achieves significantly lower validation loss compared to source-based approaches, creating a better optimization landscape for model training. We will make our code, annotated datasets, and topic classification models publicly available to facilitate further research.

cs.CL↗
arXiv2025

Vertical Federated Continual Learning via Evolving Prototype Knowledge

Vertical Federated Learning (VFL) has garnered significant attention as a privacy-preserving machine learning framework for sample-aligned feature federation. However, traditional VFL approaches do not address the challenges of class and feature continual learning, resulting in catastrophic forgetting of knowledge from previous tasks. To address the above challenge, we propose a novel vertical federated continual learning method, named Vertical Federated Continual Learning via Evolving Prototype Knowledge (V-LETO), which primarily facilitates the transfer of knowledge from previous tasks through the evolution of prototypes. Specifically, we propose an evolving prototype knowledge method, enabling the global model to retain both previous and current task knowledge. Furthermore, we introduce a model optimization technique that mitigates the forgetting of previous task knowledge by restricting updates to specific parameters of the local model, thereby enhancing overall performance. Extensive experiments conducted in both CIL and FIL settings demonstrate that our method, V-LETO, outperforms the other state-of-the-art methods. For example, our method outperforms the state-of-the-art method by 10.39% and 35.15% for CIL and FIL tasks, respectively. Our code is available at https://anonymous.4open.science/r/V-LETO-0108/README.md.

cs.LG↗
arXiv2025

Adaptive Prototype Knowledge Transfer for Federated Learning with Mixed Modalities and Heterogeneous Tasks

Multimodal Federated Learning (MFL) with mixed modalities enables unimodal and multimodal clients to collaboratively train models while ensuring clients' privacy. As a representative sample of local data, prototypes offer an approach with low resource consumption and no reliance on prior knowledge for MFL with mixed modalities. However, existing prototype-based MFL methods assume unified labels across clients and identical tasks per client, which is impractical in MFL with mixed modalities. In this work, we propose an Adaptive prototype-based Multimodal Federated Learning (AproMFL) framework for mixed modalities to address the aforementioned issues. Our AproMFL transfers knowledge through adaptively-constructed prototypes without unified labels. Clients adaptively select prototype construction methods in line with labels; server converts client prototypes into unified multimodal prototypes and cluster them to form global prototypes. To address model aggregation issues in task heterogeneity, we develop a client relationship graph-based scheme to dynamically adjust aggregation weights. Furthermore, we propose a global prototype knowledge transfer loss and a global model knowledge transfer loss to enable the transfer of global knowledge to local knowledge. Experimental results show that AproMFL outperforms four baselines on three highly heterogeneous datasets ($α=0.1$) and two heterogeneous tasks, with the optimal results in accuracy and recall being 0.42%~6.09% and 1.6%~3.89% higher than those of FedIoT (FedAvg-based MFL), respectively.

cs.LG↗
arXiv2025

RecGPT Technical Report

Recommender systems are among the most impactful applications of artificial intelligence, serving as critical infrastructure connecting users, merchants, and platforms. However, most current industrial systems remain heavily reliant on historical co-occurrence patterns and log-fitting objectives, i.e., optimizing for past user interactions without explicitly modeling user intent. This log-fitting approach often leads to overfitting to narrow historical preferences, failing to capture users' evolving and latent interests. As a result, it reinforces filter bubbles and long-tail phenomena, ultimately harming user experience and threatening the sustainability of the whole recommendation ecosystem. To address these challenges, we rethink the overall design paradigm of recommender systems and propose RecGPT, a next-generation framework that places user intent at the center of the recommendation pipeline. By integrating large language models (LLMs) into key stages of user interest mining, item retrieval, and explanation generation, RecGPT transforms log-fitting recommendation into an intent-centric process. To effectively align general-purpose LLMs to the above domain-specific recommendation tasks at scale, RecGPT incorporates a multi-stage training paradigm, which integrates reasoning-enhanced pre-alignment and self-training evolution, guided by a Human-LLM cooperative judge system. Currently, RecGPT has been fully deployed on the Taobao App. Online experiments demonstrate that RecGPT achieves consistent performance gains across stakeholders: users benefit from increased content diversity and satisfaction, merchants and the platform gain greater exposure and conversions. These comprehensive improvement results across all stakeholders validates that LLM-driven, intent-centric design can foster a more sustainable and mutually beneficial recommendation ecosystem.

cs.IR↗
arXiv2025

YOLO-PRO: Enhancing Instance-Specific Object Detection with Full-Channel Global Self-Attention

This paper addresses the inherent limitations of conventional bottleneck structures (diminished instance discriminability due to overemphasis on batch statistics) and decoupled heads (computational redundancy) in object detection frameworks by proposing two novel modules: the Instance-Specific Bottleneck with full-channel global self-attention (ISB) and the Instance-Specific Asymmetric Decoupled Head (ISADH). The ISB module innovatively reconstructs feature maps to establish an efficient full-channel global attention mechanism through synergistic fusion of batch-statistical and instance-specific features. Complementing this, the ISADH module pioneers an asymmetric decoupled architecture enabling hierarchical multi-dimensional feature integration via dual-stream batch-instance representation fusion. Extensive experiments on the MS-COCO benchmark demonstrate that the coordinated deployment of ISB and ISADH in the YOLO-PRO framework achieves state-of-the-art performance across all computational scales. Specifically, YOLO-PRO surpasses YOLOv8 by 1.0-1.6% AP (N/S/M/L/X scales) and outperforms YOLO11 by 0.1-0.5% AP in critical N/M/L/X groups, while maintaining competitive computational efficiency. This work provides practical insights for developing high-precision detectors deployable on edge devices.

cs.CV↗
arXiv2025

SciToolAgent: A Knowledge Graph-Driven Scientific Agent for Multi-Tool Integration

Scientific research increasingly relies on specialized computational tools, yet effectively utilizing these tools demands substantial domain expertise. While Large Language Models (LLMs) show promise in tool automation, they struggle to seamlessly integrate and orchestrate multiple tools for complex scientific workflows. Here, we present SciToolAgent, an LLM-powered agent that automates hundreds of scientific tools across biology, chemistry, and materials science. At its core, SciToolAgent leverages a scientific tool knowledge graph that enables intelligent tool selection and execution through graph-based retrieval-augmented generation. The agent also incorporates a comprehensive safety-checking module to ensure responsible and ethical tool usage. Extensive evaluations on a curated benchmark demonstrate that SciToolAgent significantly outperforms existing approaches. Case studies in protein engineering, chemical reactivity prediction, chemical synthesis, and metal-organic framework screening further demonstrate SciToolAgent's capability to automate complex scientific workflows, making advanced research tools accessible to both experts and non-experts.

cs.AI↗
arXiv2025

LLAMA: Multi-Feedback Smart Contract Fuzzing Framework with LLM-Guided Seed Generation

Smart contracts play a pivotal role in blockchain ecosystems, and fuzzing remains an important approach to securing smart contracts. Even though mutation scheduling is a key factor influencing fuzzing effectiveness, existing fuzzers have primarily explored seed scheduling and generation, while mutation scheduling has been rarely addressed by prior work. In this work, we propose a Large Language Models (LLMs)-based Multi-feedback Smart Contract Fuzzing framework (LLAMA) that integrates LLMs, evolutionary mutation strategies, and hybrid testing techniques. Key components of the proposed LLAMA include: (i) a hierarchical prompting strategy that guides LLMs to generate semantically valid initial seeds, coupled with a lightweight pre-fuzzing phase to select high-potential inputs; (ii) a multi-feedback optimization mechanism that simultaneously improves seed generation, seed selection, and mutation scheduling by leveraging runtime coverage and dependency feedback; and (iii) an evolutionary fuzzing engine that dynamically adjusts mutation operator probabilities based on effectiveness, while incorporating symbolic execution to escape stagnation and uncover deeper vulnerabilities. Our experiments demonstrate that LLAMA outperforms state-of-the-art fuzzers in both coverage and vulnerability detection. Specifically, it achieves 91% instruction coverage and 90% branch coverage, while detecting 132 out of 148 known vulnerabilities across diverse categories. These results highlight LLAMA's effectiveness, adaptability, and practicality in real-world smart contract security testing scenarios.

cs.SE↗