Author name cluster

Dongxiao Yu

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

19 papers

2 author rows

AAAI Conference 2026 Conference Paper

DGTF: Cross-Domain Decentralized Graph Learning with Topology-Aware Knowledge Fusion

Ruisheng Zheng
Mingyi Li
Xiao Zhang
Hongjian Shi
Yanjie Fu
Yuan Yuan
Dongxiao Yu

Cross-Domain Decentralized Graph Learning (CD-DGL) is a promising paradigm that enables efficient, privacy-preserving collaboration among multiple parties to unlock the value of cross-domain graph data. However, it faces two fundamental challenges. First, inconsistent label spaces across domains drive local models to learn domain-specific biases, which means domain-invariant topological knowledge extraction beyond label constraints is difficult. Second, existing domain topology shift and heterogeneous model architectures make direct model aggregation infeasible. To address these issues, we first use Extended Persistent Homology (EPH) to reveal and quantify the problem of domain topology shift induced by the cross-domain setting. Building on this insight, we present Decentralized Graph Learning with Topology-Aware Knowledge Fusion (DGTF), a novel framework designed to facilitate positive topological knowledge transfer in CD-DGL. Our framework achieves this by integrating two core strategies: first, a contrastive learning-based approach to extract task-agnostic topological knowledge, and second, a topology-aware, model-independent knowledge fusion method to effectively integrate this topological information. Extensive experiments conducted under various cross-domain and model-heterogeneous settings validate the superiority and effectiveness of our proposed framework.

PDF Details DOI

TCS Journal 2025 Journal Article

Adaptive pruning-based Newton's method for distributed learning

Shuzhen Chen
Yuan Yuan
Youming Tao
Tianzhu Wang
Zhipeng Cai
Dongxiao Yu

Newton's method leverages curvature information to boost performance, and thus outperforms first-order methods for distributed learning problems. However, Newton's method is not practical in large-scale and heterogeneous learning environments, due to obstacles such as high computation and communication costs of the Hessian matrix, sub-model diversity, staleness of training, and data heterogeneity. To overcome these obstacles, this paper presents a novel and efficient algorithm named Distributed Adaptive Newton Learning (DANL), which solves the drawbacks of Newton's method by using a simple Hessian initialization and adaptive allocation of training regions. The algorithm exhibits remarkable convergence properties, which are rigorously examined under standard assumptions in stochastic optimization. The theoretical analysis proves that DANL attains a linear convergence rate while efficiently adapting to available resources and keeping high efficiency. Furthermore, DANL shows notable independence from the condition number of the problem and removes the necessity for complex parameter tuning. Experiments demonstrate that DANL achieves linear convergence with efficient communication and strong performance across different datasets.

Details DOI

IJCAI Conference 2025 Conference Paper

DiffECG: Diffusion Model-Powered Label-Efficient and Personalized Arrhythmia Diagnosis

Tianren Zhou
Zhenge Jia
Dongxiao Yu
Zhaoyan Shen

Arrhythmia diagnosis using electrocardiogram (ECG) is critical for preventing cardiovascular risks. However, existing deep learning-based methods struggle with label scarcity and contrastive learning-based methods suffer from false-negative samples, which lead to poor model generalization. Besides, due to inter-subject variability, pre-trained models cannot achieve evenly performance across individuals. Conducting model fine-tuning for each individual is computationally expensive and does not guarantee improvement. We propose DiffECG, a diffusion-based self-supervised learning framework for label-efficient and personalized arrhythmia detection. Our method utilizes a diffusion model to extract robust ECG representations, coupled with a novel feature extractor and a multi-modal feature fusion strategy to obtain a well-generalized model. Moreover, we propose an efficient model personalization mechanism based on zeroth-order optimization. It personalizes the model by tuning the noise-adding step t in the diffusion process, significantly reducing computational costs compared to model fine-tuning. Experimental results show that our proposed method outperforms the SOTA method by 37. 9% and 23. 9% in generalization and personalization performance, respectively. The source code is available at: https: //github. com/Auguuust/DiffEC

PDF Details DOI

ICML Conference 2025 Conference Paper

How Distributed Collaboration Influences the Diffusion Model Training? A Theoretical Perspective

Jing Qiao
Yu Liu 0085
Yuan Yuan 0040
Xiao Zhang 0015
Zhipeng Cai 0001
Dongxiao Yu

This paper examines the theoretical performance of distributed diffusion models in environments where computational resources and data availability vary significantly among workers. Traditional models centered on single-worker scenarios fall short in such distributed settings, particularly when some workers are resource-constrained. This discrepancy in resources and data diversity challenges the assumption of accurate score function estimation foundational to single-worker models. We establish the inaugural generation error bound for distributed diffusion models in resource-limited settings, establishing a linear relationship with the data dimension $d$ and consistency with established single-worker results. Our analysis highlights the critical role of hyperparameter selection in influencing the training dynamics, which are key to the performance of model generation. This study provides a streamlined theoretical approach to optimizing distributed diffusion models, paving the way for future research in this area.

Details

ICML Conference 2025 Conference Paper

PDUDT: Provable Decentralized Unlearning under Dynamic Topologies

Jing Qiao
Yu Liu 0085
Zengzhe Chen
Mingyi Li
Yuan Yuan 0014
Xiao Zhang 0015
Dongxiao Yu

This paper investigates decentralized unlearning, aiming to eliminate the impact of a specific client on the whole decentralized system. However, decentralized communication characterizations pose new challenges for effective unlearning: the indirect connections make it difficult to trace the specific client’s impact, while the dynamic topology limits the scalability of retraining-based unlearning methods. In this paper, we propose the first P rovable D ecentralized U nlearning algorithm under D ynamic T opologies called PDUDT. It allows clients to eliminate the influence of a specific client without additional communication or retraining. We provide rigorous theoretical guarantees for PDUDT, showing it is statistically indistinguishable from perturbed retraining. Additionally, it achieves an efficient convergence rate of $\mathcal{O}(\frac{1}{T})$ in subsequent learning, where $T$ is the total communication rounds. This rate matches state-of-the-art results. Experimental results show that compared with the Retrain method, PDUDT saves more than 99% of unlearning time while achieving comparable unlearning performance.

Details

TCS Journal 2025 Journal Article

Robust matroid bandit optimization: Near-optimal rates under adversarial contamination

Youming Tao
Xiuzhen Cheng
Falko Dressler
Zhipeng Cai
Dongxiao Yu

We study the matroid bandit optimization problem, a fundamental and broadly applicable framework for combinatorial multi-armed bandits where the action space is constrained by a matroid. In particular, we address the challenge of designing algorithms that remain effective under adversarial contamination of feedback rewards, which may severely degrade performance or even mislead existing methods. Our main contribution is an efficient and robust algorithm named ROMM, which builds upon the principle of optimistic matroid maximization and leverages robust statistical estimators to assess base arm quality in polynomial time. Under the ϵ-contamination model, we establish lower bounds and prove that ROMM achieves near-optimal regret guarantees up to polylogarithmic factors. Our analysis further reveals a sharp phase transition between the low and high contamination regimes. Notably, ROMM can tolerate up to a universal constant fraction of corrupted feedback, which is optimal under mild conditions. Finally, we validate our theoretical findings with numerical experiments that demonstrate the effectiveness of the proposed method.

Details DOI

NeurIPS Conference 2025 Conference Paper

Second-Order Convergence in Private Stochastic Non-Convex Optimization

Youming Tao
Zuyuan Zhang
Dongxiao Yu
Xiuzhen Cheng
Falko Dressler
Di Wang

We investigate the problem of finding second-order stationary points (SOSP) in differentially private (DP) stochastic non-convex optimization. Existing methods suffer from two key limitations: \textbf{(i)} inaccurate convergence error rate due to overlooking gradient variance in the saddle point escape analysis, and \textbf{(ii)} dependence on auxiliary private model selection procedures for identifying DP-SOSP, which can significantly impair utility, particularly in distributed settings. To address these issues, we propose a generic perturbed stochastic gradient descent (PSGD) framework built upon Gaussian noise injection and general gradient oracles. A core innovation of our framework is using model drift distance to determine whether PSGD escapes saddle points, ensuring convergence to approximate local minima without relying on second-order information or additional DP-SOSP identification. By leveraging the adaptive DP-SPIDER estimator as a specific gradient oracle, we develop a new DP algorithm that rectifies the convergence error rates reported in prior work. We further extend this algorithm to distributed learning with heterogeneous data, providing the first formal guarantees for finding DP-SOSP in such settings. Our analysis also highlights the detrimental impacts of private selection procedures in distributed learning under high-dimensional models, underscoring the practical benefits of our design. Numerical experiments on real-world datasets validate the efficacy of our approach.

PDF Details

AAAI Conference 2024 Conference Paper

ConcaveQ: Non-monotonic Value Function Factorization via Concave Representations in Deep Multi-Agent Reinforcement Learning

Huiqun Li
Hanhan Zhou
Yifei Zou
Dongxiao Yu
Tian Lan

Value function factorization has achieved great success in multi-agent reinforcement learning by optimizing joint action-value functions through the maximization of factorized per-agent utilities. To ensure Individual-Global-Maximum property, existing works often focus on value factorization using monotonic functions, which are known to result in restricted representation expressiveness. In this paper, we analyze the limitations of monotonic factorization and present ConcaveQ, a novel non-monotonic value function factorization approach that goes beyond monotonic mixing functions and employs neural network representations of concave mixing functions. Leveraging the concave property in factorization, an iterative action selection scheme is developed to obtain optimal joint actions during training. It is used to update agents’ local policy networks, enabling fully decentralized execution. The effectiveness of the proposed ConcaveQ is validated across scenarios involving multi-agent predator-prey environment and StarCraft II micromanagement tasks. Empirical results exhibit significant improvement of ConcaveQ over state-of-the-art multi-agent reinforcement learning approaches.

PDF Details DOI

ECAI Conference 2024 Conference Paper

Federation-Paced Learning: Towards Efficient Federated Learning with Synchronized Pace

Tingting Zhang
Mei Cao
Zhenge Jia
Jianbo Lu
Zhaoyan Shen
Dongxiao Yu
Mengying Zhao

Federated learning (FL) is a distributed machine learning approach that allows multiple devices or computing nodes to jointly train models without sharing raw data. However, in real-world application scenarios, FL usually encounters a critical challenge of data heterogeneity. Recent studies have revealed that the client’s model suffers severe bias between the local model and global model, leading to global performance degradation. Improving the generalization of local learning would inherently reduce bias. It has been proved that self-paced learning on a single device can greatly achieve a better generalization result. However, it is not well explored how it can be applied to federated learning with a number of distributed nodes working cooperatively. Specifically, self-paced learning suggests using easy data and then gradually difficult data during model training. It is not straightforward to differentiate “easy” and “difficult” data at the local since global data distribution is not available, especially with severe data heterogeneity. To address the above issues, we propose a novel federated learning framework, Federation-Paced Learning (FedPL), which enables a self-paced process in federated learning and effectively improves the model performance. First, we propose schemes to analyze the data characteristics in terms of difficulty. Then we define a stage controller to synchronize the learning process across cooperative nodes to follow the easy-to-hard rule. Finally, we propose a client selection strategy to further improve the learning efficacy. We evaluate the performance of FedPL on several generic public datasets. Experiment results show that the proposed FedPL outperforms existing methods by up to 13. 50% in terms of accuracy. Code is available at https: //github. com/tnghua/FedPL.

Details

AAAI Conference 2024 Conference Paper

patchDPCC: A Patchwise Deep Compression Framework for Dynamic Point Clouds

Zirui Pan
Mengbai Xiao
Xu Han
Dongxiao Yu
Guanghui Zhang
Yao Liu

When compressing point clouds, point-based deep learning models operate points in a continuous space, which has a chance to minimize the geometric fidelity loss introduced by voxelization in preprocessing. But these methods could hardly scale to inputs with arbitrary points. Furthermore, the point cloud frames are individually compressed, failing the conventional wisdom of leveraging inter-frame similarity. In this work, we propose a patchwise compression framework called patchDPCC, which consists of a patch group generation module and a point-based compression model. Algorithms are developed to generate patches from different frames representing the same object, and more importantly, these patches are regulated to have the same number of points. We also incorporate a feature transfer module in the compression model, which refines the feature quality by exploiting the inter-frame similarity. Our model generates point-wise features for entropy coding, which guarantees the reconstruction speed. The evaluation on the MPEG 8i dataset shows that our method improves the compression ratio by 47.01% and 85.22% when compared to PCGCv2 and V-PCC with the same reconstruction quality, which is 9% and 16% better than that D-DPCC does. Our method also achieves the fastest decoding speed among the learning-based compression models.

PDF Details DOI

NeurIPS Conference 2024 Conference Paper

Resource-Aware Federated Self-Supervised Learning with Global Class Representations

Mingyi Li
Xiao Zhang
Qi Wang
Tengfei Liu
Ruofan Wu
Weiqiang Wang
Fuzhen Zhuang
Hui Xiong

Due to the heterogeneous architectures and class skew, the global representation models training in resource-adaptive federated self-supervised learning face with tricky challenges: $\textit{deviated representation abilities}$ and $\textit{inconsistent representation spaces}$. In this work, we are the first to propose a multi-teacher knowledge distillation framework, namely $\textit{FedMKD}$, to learn global representations with whole class knowledge from heterogeneous clients even under extreme class skew. Firstly, the adaptive knowledge integration mechanism is designed to learn better representations from all heterogeneous models with deviated representation abilities. Then the weighted combination of the self-supervised loss and the distillation loss can support the global model to encode all classes from clients into a unified space. Besides, the global knowledge anchored alignment module can make the local representation spaces close to the global spaces, which further improves the representation abilities of local ones. Finally, extensive experiments conducted on two datasets demonstrate the effectiveness of $\textit{FedMKD}$ which outperforms state-of-the-art baselines 4. 78\% under linear evaluation on average.

PDF Details DOI

JBHI Journal 2022 Journal Article

HarMI: Human Activity Recognition Via Multi-Modality Incremental Learning

Xiao Zhang
Hongzheng Yu
Yang Yang
Jingjing Gu
Yujun Li
Fuzhen Zhuang
Dongxiao Yu
Zhaochun Ren

Nowadays, with the development of various kinds of sensors in smartphones or wearable devices, human activity recognition (HAR) has been widely researched and has numerous applications in healthcare, smart city, etc. Many techniques based on hand-crafted feature engineering or deep neural network have been proposed for sensor based HAR. However, these existing methods usually recognize activities offline, which means the whole data should be collected before training, occupying large-capacity storage space. Moreover, once the offline model training finished, the trained model can’t recognize new activities unless retraining from the start, thus with a high cost of time and space. In this paper, we propose a multi-modality incremental learning model, called HarMI, with continuous learning ability. The proposed HarMI model can start training quickly with little storage space and easily learn new activities without storing previous training data. In detail, we first adopt attention mechanism to align heterogeneous sensor data with different frequencies. In addition, to overcome catastrophic forgetting in incremental learning, HarMI utilizes the elastic weight consolidation and canonical correlation analysis from a multi-modality perspective. Extensive experiments based on two public datasets demonstrate that HarMI can achieve a superior performance compared with several state-of-the-arts.

Details DOI

TCS Journal 2020 Journal Article

Offline and online algorithms for single-minded selling problem

Yong Zhang
Francis Y.L. Chin
Sheung-Hung Poon
Hing-Fung Ting
Dachuan Xu
Dongxiao Yu

Given a seller with k types of items and n single-minded buyers, i. e. , each buyer is only interested in a particular bundle of items, to maximize the revenue, the seller must assign some amount of bundles to each buyer with respect to the buyer's accepted price. Each buyer b i is associated with a value function v i ( ⋅ ) such that v i ( x ) is the accepted unit bundle price b i is willing to pay for x bundles. In this paper, we assume that bundles can be sold fractionally. The single-minded item selling problem is proved to be NP-hard. Moreover, we give an O ( k ) -approximation algorithm. For the online version, i. e. , buyers come one by one and the decision must be made immediately on the arrival of each buyer, an O ( k ⋅ ( log ⁡ h + log ⁡ k ) ) -competitive algorithm is given, where h is the highest unit item price among all buyers.

Details DOI

AAMAS Conference 2017 Conference Paper

Uniform Information Exchange in Multi-channel Wireless Ad Hoc Networks

Dongxiao Yu
Li Ning
Yong Zhang
Hai Jin
Yuexuan Wang
Francis C. M. Lau
Shengzhong Feng

Information exchange is a basic primitive for maintaining the smooth running of a network or a system with multiple communicating agents. Given k packets initially stored at k nodes respectively, the problem is to disseminate the k packets to the whole network with the objective of minimizing the time used. We study this problem in single-hop multi-channel networks of n nodes, and target on devising uniform distributed protocols that do not rely on any prior knowledge of network parameters, such as the network size n or the number of packet holders k. Uniform protocols have better scalability and are more suitable for implementation in reality. Specifically, we propose a uniform distributed protocol that with high probability accomplishes the dissemination in O(k/F + F · log n) rounds, assuming F available channels. This protocol is asymptotically optimal when k is large (k ≥ F2 · log n), and provides the best possible linear speedup with multiple channels comparing with the results using a single channel. To the best of our knowledge, this is the first uniform protocol for information exchange in multichannel networks.

PDF

TCS Journal 2016 Journal Article

Distributed multiple-message broadcast in wireless ad hoc networks under the SINR model

Dongxiao Yu
Qiang-Sheng Hua
Yuexuan Wang
Haisheng Tan
Francis C.M. Lau

In a multiple-message broadcast, an arbitrary number of messages originate at arbitrary nodes in the network at arbitrary times. The problem is to disseminate all these messages to the whole network. This paper gives the first randomized distributed multiple-message broadcast algorithm with worst-case performance guarantee in wireless ad hoc networks employing the SINR interference model which takes interferences from all the nodes in the network into account. The network model used in this paper also considers the harsh characteristics of wireless ad hoc networks: there is no prior structure, and nodes cannot perform collision detection and have little knowledge of the network topology. Under all these restrictions, our proposed randomized distributed multiple-message broadcast protocol can deliver any message m to all nodes in the network in O ( D + k + log 2 ⁡ n ) timeslots with high probability, where D is the network diameter, k is the number of messages whose broadcasts overlap with m, and n is the number of nodes in the network. We also study the lower bound for randomized distributed multiple-message broadcast protocols. In particular, we prove that any uniform randomized algorithm needs Ω ( D + k + log 2 ⁡ n log ⁡ log ⁡ log ⁡ n ) timeslots to disseminate k messages initially stored at k nodes.

Details DOI

TCS Journal 2014 Journal Article

Distributed ( Δ + 1 ) -coloring in the physical model

Dongxiao Yu
Yuexuan Wang
Qiang-Sheng Hua
Francis C.M. Lau

In multi-hop radio networks, such as wireless ad-hoc networks and wireless sensor networks, nodes employ a MAC (Medium Access Control) protocol such as TDMA to coordinate accesses to the shared medium and to avoid interference of close-by transmissions. These protocols can be implemented using standard node coloring. The ( Δ + 1 ) -coloring problem is to color all nodes in as few timeslots as possible using at most Δ + 1 colors such that any two nodes within distance R are assigned different colors, where R is a given parameter and Δ is the maximum degree of the modeled unit disk graph using R as a scaling factor. Being one of the most fundamental problems in distributed computing, this problem is well studied and there is a long chain of algorithms prescribed for it. However, all previous works are based on abstract models, such as message passing models and graph based interference models, which limit the utility of these algorithms in practice. In this paper, for the first time, we consider the distributed ( Δ + 1 ) -coloring problem under the more practical SINR interference model. In particular, without requiring any knowledge about the neighborhood, we propose a novel randomized ( Δ + 1 ) -coloring algorithm with time complexity O ( Δ log ⁡ n + log 2 ⁡ n ). For the case where nodes cannot adjust their transmission power, we give an O ( Δ log 2 ⁡ n ) randomized algorithm, which only incurs a logarithmic multiplicative factor overhead.

Details DOI

TCS Journal 2010 Journal Article

Dynamic programming based algorithms for set multicover and multiset multicover problems

Qiang-Sheng Hua
Yuexuan Wang
Dongxiao Yu
Francis C.M. Lau

Given a universe N containing n elements and a collection of multisets or sets over N, the multiset multicover (MSMC) problem or the set multicover (SMC) problem is to cover all elements at least a number of times as specified in their coverage requirements with the minimum number of multisets or sets. In this paper, we give various exact algorithms for these two problems with or without constraints on the number of times a multiset or set may be chosen. First, we show that the MSMC without multiplicity constraints problem can be solved in O ∗ ( ( b + 1 ) n | F | ) time and polynomial space, where b is the maximum coverage requirement and | F | denotes the total number of given multisets over N. (The O ∗ notation suppresses a factor polynomial in n.) To our knowledge, this is the first known exact algorithm for the MSMC without multiplicity constraints problem. Second, by combining dynamic programming and the inclusion–exclusion principle, we can exactly solve the SMC without multiplicity constraints problem in O ( ( b + 2 ) n ) time. Compared with two recent results, in [Q. -S. Hua, Y. Wang, D. Yu, F. C. M. Lau, Set multi-covering via inclusion–exclusion, Theoretical Computer Science, 410 (38–40) (2009) 3882–3892] and [J. Nederlof, Inclusion exclusion for hard problems, Master Thesis, Utrecht University, The Netherlands, 2008], respectively, ours is the fastest exact algorithm for the SMC without multiplicity constraints problem. Finally, by directly using dynamic programming, we give the first known exact algorithm for the MSMC or the SMC with multiplicity constraints problem in O ( ( b + 1 ) n | F | ) time and O ∗ ( ( b + 1 ) n ) space. This algorithm can also be easily adapted as a constructive algorithm for the MSMC without multiplicity constraints problem.

Details DOI

TCS Journal 2009 Journal Article

Acyclic edge coloring of planar graphs with large girth

Dongxiao Yu
Jianfeng Hou
Guizhen Liu
Bin Liu
Lan Xu

Acyclic coloring problem is a specialized problem that arises in the efficient computation of Hessians. A proper edge coloring of a graph G is called acyclic if there is no 2 -colored cycle in G. The acyclic edge chromatic number χ a ′ ( G ) of G is the least number of colors in an acyclic edge coloring of G. Alon et al. conjectured that χ a ′ ( G ) ≤ Δ ( G ) + 2. In this paper, we consider the sufficient conditions for the planar graphs satisfying χ a ′ ( G ) ≤ Δ ( G ) + 1 and χ a ′ ( G ) = Δ ( G ).

Details DOI

TCS Journal 2009 Journal Article

Set multi-covering via inclusion–exclusion

Qiang-Sheng Hua
Yuexuan Wang
Dongxiao Yu
Francis C.M. Lau

Set multi-covering is a generalization of the set covering problem where each element may need to be covered more than once and thus some subset in the given family of subsets may be picked several times for minimizing the number of sets to satisfy the coverage requirement. In this paper, we propose a family of exact algorithms for the set multi-covering problem based on the inclusion–exclusion principle. The presented ESMC (Exact Set Multi-Covering) algorithm takes O ∗ ( ( 2 t ) n ) time and O ∗ ( ( t + 1 ) n ) space where t is the maximum value in the coverage requirement set (The O ∗ ( f ( n ) ) notation omits a p o l y log ( f ( n ) ) factor). We also propose the other three exact algorithms through different tradeoffs of the time and space complexities. To the best of our knowledge, this present paper is the first one to give exact algorithms for the set multi-covering problem with nontrivial time and space complexities. This paper can also be regarded as a generalization of the exact algorithm for the set covering problem given in [A. Björklund, T. Husfeldt, M. Koivisto, Set partitioning via inclusion–exclusion, SIAM Journal on Computing, in: FOCS 2006 (in press, special issue)].

Details DOI