Sample Condensation in Online Continual Learning

TL;DR

This paper introduces OLCGM, a novel replay-based continual learning method that uses knowledge condensation to compress memory samples, improving accuracy in scenarios with limited memory capacity.

Contribution

OLCGM is the first to apply knowledge condensation for memory compression in online continual learning, enhancing performance with limited memory.

Findings

OLCGM outperforms state-of-the-art replay strategies under limited memory conditions.

Memory condensation maintains higher sample quality over time.

OLCGM achieves better accuracy when data complexity exceeds memory size.

Abstract

Online Continual learning is a challenging learning scenario where the model must learn from a non-stationary stream of data where each sample is seen only once. The main challenge is to incrementally learn while avoiding catastrophic forgetting, namely the problem of forgetting previously acquired knowledge while learning from new data. A popular solution in these scenario is to use a small memory to retain old data and rehearse them over time. Unfortunately, due to the limited memory size, the quality of the memory will deteriorate over time. In this paper we propose OLCGM, a novel replay-based continual learning strategy that uses knowledge condensation techniques to continuously compress the memory and achieve a better use of its limited size. The sample condensation step compresses old samples, instead of removing them like other replay strategies. As a result, the experiments show that, whenever the memory budget is limited compared to the complexity of the data, OLCGM improves the final accuracy compared to state-of-the-art replay strategies.