Generalizable Heterogeneous Federated Cross-Correlation and Instance Similarity Learning

TL;DR

This paper introduces FCCL+, a federated learning method that enhances model generalization and mitigates catastrophic forgetting by leveraging cross-correlation, instance similarity, and non-target distillation, validated through a comprehensive benchmark.

Contribution

The paper proposes FCCL+, a novel federated learning framework that improves heterogeneity handling and generalization using cross-correlation and non-target distillation techniques.

Findings

FCCL+ outperforms existing methods across multiple domain shift scenarios.

The use of cross-correlation matrices enhances intra-domain discriminability.

Federated Non Target Distillation effectively mitigates catastrophic forgetting.

Abstract

Federated learning is an important privacy-preserving multi-party learning paradigm, involving collaborative learning with others and local updating on private data. Model heterogeneity and catastrophic forgetting are two crucial challenges, which greatly limit the applicability and generalizability. This paper presents a novel FCCL+, federated correlation and similarity learning with non-target distillation, facilitating the both intra-domain discriminability and inter-domain generalization. For heterogeneity issue, we leverage irrelevant unlabeled public data for communication between the heterogeneous participants. We construct cross-correlation matrix and align instance similarity distribution on both logits and feature levels, which effectively overcomes the communication barrier and improves the generalizable ability. For catastrophic forgetting in local updating stage, FCCL+ introduces Federated Non Target Distillation, which retains inter-domain knowledge while avoiding the optimization conflict issue, fulling distilling privileged inter-domain information through depicting posterior classes relation. Considering that there is no standard benchmark for evaluating existing heterogeneous federated learning under the same setting, we present a comprehensive benchmark with extensive representative methods under four domain shift scenarios, supporting both heterogeneous and homogeneous federated settings. Empirical results demonstrate the superiority of our method and the efficiency of modules on various scenarios.