A Comprehensive Empirical Evaluation on Online Continual Learning

TL;DR

This paper provides an extensive empirical comparison of online continual learning methods for image classification, highlighting stability issues and the strength of simple replay baselines, with insights into representation quality and practical evaluation metrics.

Contribution

It offers a comprehensive evaluation of existing online continual learning methods on standard benchmarks, revealing stability challenges and emphasizing the effectiveness of well-tuned replay baselines.

Findings

Most methods face stability and underfitting issues.

Learned representations are comparable to i.i.d. training.

Basic experience replay is a strong baseline when properly tuned.

Abstract

Online continual learning aims to get closer to a live learning experience by learning directly on a stream of data with temporally shifting distribution and by storing a minimum amount of data from that stream. In this empirical evaluation, we evaluate various methods from the literature that tackle online continual learning. More specifically, we focus on the class-incremental setting in the context of image classification, where the learner must learn new classes incrementally from a stream of data. We compare these methods on the Split-CIFAR100 and Split-TinyImagenet benchmarks, and measure their average accuracy, forgetting, stability, and quality of the representations, to evaluate various aspects of the algorithm at the end but also during the whole training period. We find that most methods suffer from stability and underfitting issues. However, the learned representations are comparable to i.i.d. training under the same computational budget. No clear winner emerges from the results and basic experience replay, when properly tuned and implemented, is a very strong baseline. We release our modular and extensible codebase at https://github.com/AlbinSou/ocl_survey based on the avalanche framework to reproduce our results and encourage future research.