A Game-Theoretic Approach for Robust Federated Learning

  • سال انتشار: 1400
  • محل انتشار: ماهنامه بین المللی مهندسی، دوره: 34، شماره: 4
  • کد COI اختصاصی: JR_IJE-34-4_009
  • زبان مقاله: انگلیسی
  • تعداد مشاهده: 557
دانلود فایل این مقاله

نویسندگان

E. Tahanian

Faculty of Computer Engineering, Shahrood University of Technology, Shahrood, Iran

M. Amouei

Faculty of Computer Engineering, Shahrood University of Technology, Shahrood, Iran

H. Fateh

Faculty of Computer Engineering, Shahrood University of Technology, Shahrood, Iran

M. Rezvani

Faculty of Computer Engineering, Shahrood University of Technology, Shahrood, Iran

چکیده

Federated Learning enables aggregating models trained over a large number of clients by sending these models to a central server, while data privacy is preserved since only the models are sent. Federated learning techniques are considerably vulnerable to poisoning attacks. In this paper, we explore the threat of poisoning attacks and introduce a game-based robust federated averaging algorithm to detect and discard bad updates provided by the clients. We model the aggregating process with a mixed-strategy game that is played between the server and each client. The valid actions of the clients are to send good or bad updates while the server can accept or ignore these updates as its valid actions. By employing the Nash Equilibrium property, the server determines the probability of providing good updates by each client. The experimental results show that our proposed game-based aggregation algorithm is significantly more robust to faulty and noisy clients in comparison with the most recently presented methods. According to these results, our algorithm converges after a maximum of ۳۰ iterations and can detect ۱۰۰% of the bad clients for all the investigated scenarios. In addition, the accuracy of the proposed algorithm is at least ۱۵.۸% and ۲.۳% better than state of the art for flipping and noisy scenarios, respectively.

کلیدواژه ها

Federated Learning, game theory, byzantine model, Adaptive averaging

اطلاعات بیشتر در مورد COI

COI مخفف عبارت CIVILICA Object Identifier به معنی شناسه سیویلیکا برای اسناد است. COI کدی است که مطابق محل انتشار، به مقالات کنفرانسها و ژورنالهای داخل کشور به هنگام نمایه سازی بر روی پایگاه استنادی سیویلیکا اختصاص می یابد.

کد COI به مفهوم کد ملی اسناد نمایه شده در سیویلیکا است و کدی یکتا و ثابت است و به همین دلیل همواره قابلیت استناد و پیگیری دارد.