Federated unsupervised representation learning

• Published in: Frontiers of information technology & electronic engineering Vol. 24; no. 8; pp. 1181 - 1193
• Main Authors: Zhang, Fengda, Kuang, Kun, Chen, Long, You, Zhaoyang, Shen, Tao, Xiao, Jun, Zhang, Yin, Wu, Chao, Wu, Fei, Zhuang, Yueting, Li, Xiaolin
• Format: Journal Article
• Language: English
• Published: Hangzhou Zhejiang University Press 01-08-2023; Springer Nature B.V; College of Computer Science and Technology,Zhejiang University,Hangzhou 310027,China%School of Public Affairs,Zhejiang University,Hangzhou 310027,China%Tongdun Technology,Hangzhou 310000,China; ElasticMind.AI Technology Inc.,Hangzhou 310018,China; Institute of Basic Medicine and Cancer,Chinese Academy of Sciences,Hangzhou 310018,China
• Subjects: Algorithms; Alignment; Clients; Communications Engineering; Computer Hardware; Computer Science; Computer Systems Organization and Communication Networks; Dictionaries; Electrical Engineering; Electronics and Microelectronics; Instrumentation; Modules; Networks; Representations; Research Article; Representation learning; Federated learning; 无监督学习; TP183; Contrastive learning; 表示学习; 对比学习; 联邦学习; Unsupervised learning
• ISSN: 2095-9184; 2095-9230
• DOI: 10.1631/FITEE.2200268

To leverage the enormous amount of unlabeled data on distributed edge devices, we formulate a new problem in federated learning called federated unsupervised representation learning (FURL) to learn a common representation model without supervision while preserving data privacy. FURL poses two new challenges: (1) data distribution shift (non-independent and identically distributed, non-IID) among clients would make local models focus on different categories, leading to the inconsistency of representation spaces; (2) without unified information among the clients in FURL, the representations across clients would be misaligned. To address these challenges, we propose the federated contrastive averaging with dictionary and alignment (FedCA) algorithm. FedCA is composed of two key modules: a dictionary module to aggregate the representations of samples from each client which can be shared with all clients for consistency of representation space and an alignment module to align the representation of each client on a base model trained on public data. We adopt the contrastive approach for local model training. Through extensive experiments with three evaluation protocols in IID and non-IID settings, we demonstrate that FedCA outperforms all baselines with significant margins.