Jointly Exploring Client Drift and Catastrophic Forgetting in Dynamic Learning

TL;DR

This paper introduces a unified framework to analyze the combined effects of Client Drift and Catastrophic Forgetting in dynamic learning environments, revealing their correlation and potential for improved generalization.

Contribution

It presents a novel combined analysis framework and 3D performance landscape for jointly studying Client Drift and Catastrophic Forgetting, linking their impacts and behaviors.

Findings

Performance drop from Client Drift correlates with that from Catastrophic Forgetting (Pearson > 0.94)

Moderate combined shifts can lead to a performance improvement ('Generalization Bump')

The framework enables analysis of mixed distribution shifts in federated and continual learning

Abstract

Federated and Continual Learning have emerged as potential paradigms for the robust and privacy-aware use of Deep Learning in dynamic environments. However, Client Drift and Catastrophic Forgetting are fundamental obstacles to guaranteeing consistent performance. Existing work only addresses these problems separately, which neglects the fact that the root cause behind both forms of performance deterioration is connected. We propose a unified analysis framework for building a controlled test environment for Client Drift -- by perturbing a defined ratio of clients -- and Catastrophic Forgetting -- by shifting all clients with a particular strength. Our framework further leverages this new combined analysis by generating a 3D landscape of the combined performance impact from both. We demonstrate that the performance drop through Client Drift, caused by a certain share of shifted clients, is correlated to the drop from Catastrophic Forgetting resulting from a corresponding shift strength. Correlation tests between both problems for Computer Vision (CelebA) and Medical Imaging (PESO) support this new perspective, with an average Pearson rank correlation coefficient of over 0.94. Our framework's novel ability of combined spatio-temporal shift analysis allows us to investigate how both forms of distribution shift behave in mixed scenarios, opening a new pathway for better generalization. We show that a combination of moderate Client Drift and Catastrophic Forgetting can even improve the performance of the resulting model (causing a "Generalization Bump") compared to when only one of the shifts occurs individually. We apply a simple and commonly used method from Continual Learning in the federated setting and observe this phenomenon to be reoccurring, leveraging the ability of our framework to analyze existing and novel methods for Federated and Continual Learning.