App. Res. of Comp. Accepted Paper
Vol. 43, 2026 No. 5

FedAGen: federated learning drift calibration algorithm based on adversarial pseudo-sample generation

Huang Junhuang1a,1b
Ning Jianting1a,1b
Hou Linshan2
Chen Zhihao1a,1b
Ma Chuyang3
1. a. College of Computer and Cyber Security, b. Fujian Provincial Key Laboratory of Network Security and Cryptology, Fujian Normal University, Fuzhou Fujian 350117, China
2. School of Computer Science and Technology, Harbin Institute of Technology(Shenzhen), Shenzhen 518055, China
3. College of Computer Science and Technology, Jiangsu Normal University, Xuzhou Jiangsu 221116, China
DOI: 10.19734/j.issn.1001-3695.2025.08.0399

Abstract

To address the "client drift" problem in federated learning, where local models deviate from the global optimum when trained on non-independent and identically distributed (Non-IID) data, leading to decreased accuracy and slow convergence of the global model, this paper proposes a federated learning drift calibration algorithm named FedAGen. This algorithm operates on the client side, employing a novel adversarial optimization method. Guided by the global model, it applies perturbations to local samples to generate pseudo-samples in the feature space that can simulate the global data distribution. By incorporating these pseudo-samples into local training, the algorithm proactively calibrates the model drift caused by data skew at its source. Experimental results on the CIFAR-10, CIFAR-100, and FEMNIST datasets demonstrate the superior performance of the proposed algorithm. In a Non-IID environment (Dirichlet parameter α=0.3) , the model achieves final test accuracies of 86.85%, 62.31%, and 86.02%, respectively. Compared to the classic FedAvg algorithm, FedAGen improves the accuracy on CIFAR-100 by over 11 percentage points. Furthermore, the algorithm exhibits strong robustness under varying degrees of data heterogeneity and achieves a faster convergence speed. The study shows that FedAGen can effectively mitigate client drift. Without significantly increasing communication overhead, it markedly enhances the model accuracy and training efficiency of federated learning systems in Non-IID scenarios, thus validating the effectiveness and superiority of the proposed method.

Foundation Support

国家自然科学基金资助项目(62372108,62425205)

Publish Information

DOI: 10.19734/j.issn.1001-3695.2025.08.0399
Publish at: Application Research of Computers Accepted Paper, Vol. 43, 2026 No. 5

Publish History

[2026-01-20] Accepted Paper

Cite This Article

黄俊煌, 宁建廷, 侯琳珊, 等. FedAGen：一种基于对抗性伪样本生成的联邦学习漂移校准算法 [J]. 计算机应用研究, 2026, 43 (5). (2026-01-20). https://doi.org/10.19734/j.issn.1001-3695.2025.08.0399. (Huang Junhuang, Ning Jianting, Hou Linshan, et al. FedAGen: federated learning drift calibration algorithm based on adversarial pseudo-sample generation [J]. Application Research of Computers, 2026, 43 (5). (2026-01-20). https://doi.org/10.19734/j.issn.1001-3695.2025.08.0399. )

About the Journal

  • Application Research of Computers Monthly Journal
  • Journal ID ISSN 1001-3695
    CN  51-1196/TP

Application Research of Computers, founded in 1984, is an academic journal of computing technology sponsored by Sichuan Institute of Computer Sciences under the Science and Technology Department of Sichuan Province.

Aiming at the urgently needed cutting-edge technology in this discipline, Application Research of Computers reflects the mainstream technology, hot technology and the latest development trend of computer application research at home and abroad in a timely manner. The main contents of the journal include high-level academic papers in this discipline, the latest scientific research results and major application results. The contents of the columns involve new theories of computer discipline, basic computer theory, algorithm theory research, algorithm design and analysis, blockchain technology, system software and software engineering technology, pattern recognition and artificial intelligence, architecture, advanced computing, parallel processing, database technology, computer network and communication technology, information security technology, computer image graphics and its latest hot application technology.

Application Research of Computers has many high-level readers and authors, and its readers are mainly senior and middle-level researchers and engineers engaged in the field of computer science, as well as teachers and students majoring in computer science and related majors in colleges and universities. Over the years, the total citation frequency and Web download rate of Application Research of Computers have been ranked among the top of similar academic journals in this discipline, and the academic papers published are highly popular among the readers for their novelty, academics, foresight, orientation and practicality.

Indexed & Evaluation

  • The Second National Periodical Award 100 Key Journals
  • Double Effect Journal of China Journal Formation
  • the Core Journal of China (Peking University 2023 Edition)
  • the Core Journal for Science
  • Chinese Science Citation Database (CSCD) Source Journals
  • RCCSE Chinese Core Academic Journals
  • Journal of China Computer Federation
  • 2020-2022 The World Journal Clout Index (WJCI) Report of Scientific and Technological Periodicals
  • Full-text Source Journal of China Science and Technology Periodicals Database
  • Source Journal of China Academic Journals Comprehensive Evaluation Database
  • Source Journals of China Academic Journals (CD-ROM Version), China Journal Network
  • 2017-2019 China Outstanding Academic Journals with International Influence (Natural Science and Engineering Technology)
  • Source Journal of Top Academic Papers (F5000) Program of China's Excellent Science and Technology Journals
  • Source Journal of China Engineering Technology Electronic Information Network and Electronic Technology Literature Database
  • Source Journal of British Science Digest (INSPEC)
  • Japan Science and Technology Agency (JST) Source Journal
  • Russian Journal of Abstracts (AJ, VINITI) Source Journals
  • Full-text Journal of EBSCO, USA
  • Cambridge Scientific Abstracts (Natural Sciences) (CSA(NS)) core journals
  • Poland Copernicus Index (IC)
  • Ulrichsweb (USA)