L-DAWA: Layer-wise Divergence Aware Weight Aggregation in Federated Self-Supervised Visual Representation Learning

2023/07/14 by Yasar Abbas Ur Rehman, Rehman, Yasar Abbas Ur, Yan Gao +9 · 8 citations
Computer Science · #Advanced Neural Network Applications #Computer Vision and Pattern Recognition (cs.CV) #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Privacy-Preserving Technologies in Data

paper · pdf · doi:10.48550/arxiv.2307.07393

openalex publication_date 2023/07/14 · openalex created_date 2023/07/18 · openalex updated_date 2026/07/28

Abstract

The ubiquity of camera-enabled devices has led to large amounts of unlabeled image data being produced at the edge. The integration of self-supervised learning (SSL) and federated learning (FL) into one coherent system can potentially offer data privacy guarantees while also advancing the quality and robustness of the learned visual representations without needing to move data around. However, client bias and divergence during FL aggregation caused by data heterogeneity limits the performance of learned visual representations on downstream tasks. In this paper, we propose a new aggregation strategy termed Layer-wise Divergence Aware Weight Aggregation (L-DAWA) to mitigate the influence of client bias and divergence during FL aggregation. The proposed method aggregates weights at the layer-level according to the measure of angular divergence between the clients' model and the global model. Extensive experiments with cross-silo and cross-device settings on CIFAR-10/100 and Tiny ImageNet datasets demonstrate that our methods are effective and obtain new SOTA performance on both contrastive and non-contrastive SSL approaches.

Cited by

• Towards Heterogeneous Quantum Federated Learning: Challenges and Solutions
• CATCHFed: Efficient Unlabeled Data Utilization for Semi-Supervised Federated Learning in Limited Labels Environments
• Research in Collaborative Learning Does Not Serve Cross-Silo Federated Learning in Practice
• Cooperative Pseudo Labeling for Unsupervised Federated Classification
• Foundational Models and Federated Learning: Survey, Taxonomy, Challenges and Practical Insights
• IPBA: Imperceptible Perturbation Backdoor Attack in Federated Self-Supervised Learning
• Soft Separation and Distillation: Toward Global Uniformity in Federated Unsupervised Learning
• FedCTTA: A Collaborative Approach to Continual Test-Time Adaptation in Federated Learning
• FSSUAVL: A Discriminative Framework using Vision Models for Federated Self-Supervised Audio and Image Understanding

Related

• Divergence-aware Federated Self-Supervised Learning
• Collaborative Unsupervised Visual Representation Learning from Decentralized Data
• Enabling Quartile-based Estimated-Mean Gradient Aggregation As Baseline for Federated Image Classifications
• Overcoming label shift with target-aware federated learning
• Federated Learning for Commercial Image Sources