Looking through the past: better knowledge retention for generative replay in continual learning

TL;DR

This paper enhances generative replay in continual learning by introducing three modifications that improve data generation and feature alignment, leading to better performance on complex datasets.

Contribution

It proposes a novel approach with latent space distillation, latent matching, and cycling generations to improve generative replay in complex scenarios.

Findings

Outperforms existing generative replay methods on various benchmarks.

Improves feature alignment and data quality in VAE-based generative models.

Enhances knowledge retention in continual learning settings.

Abstract

In this work, we improve the generative replay in a continual learning setting to perform well on challenging scenarios. Current generative rehearsal methods are usually benchmarked on small and simple datasets as they are not powerful enough to generate more complex data with a greater number of classes. We notice that in VAE-based generative replay, this could be attributed to the fact that the generated features are far from the original ones when mapped to the latent space. Therefore, we propose three modifications that allow the model to learn and generate complex data. More specifically, we incorporate the distillation in latent space between the current and previous models to reduce feature drift. Additionally, a latent matching for the reconstruction and original data is proposed to improve generated features alignment. Further, based on the observation that the reconstructions are better for preserving knowledge, we add the cycling of generations through the previously trained model to make them closer to the original data. Our method outperforms other generative replay methods in various scenarios. Code available at https://github.com/valeriya-khan/looking-through-the-past.