t-DGR: A Trajectory-Based Deep Generative Replay Method for Continual Learning in Decision Making

TL;DR

This paper introduces t-DGR, a non-autoregressive generative replay method for continual learning in decision-making, which effectively mitigates error accumulation and achieves state-of-the-art results on benchmark tasks.

Contribution

The paper presents a scalable, non-autoregressive generative replay approach that improves continual learning performance in decision-making tasks.

Findings

Achieves state-of-the-art success rates on Continual World benchmarks.

Outperforms autoregressive models in generating task trajectories.

Reduces error accumulation in trajectory generation.

Abstract

Deep generative replay has emerged as a promising approach for continual learning in decision-making tasks. This approach addresses the problem of catastrophic forgetting by leveraging the generation of trajectories from previously encountered tasks to augment the current dataset. However, existing deep generative replay methods for continual learning rely on autoregressive models, which suffer from compounding errors in the generated trajectories. In this paper, we propose a simple, scalable, and non-autoregressive method for continual learning in decision-making tasks using a generative model that generates task samples conditioned on the trajectory timestep. We evaluate our method on Continual World benchmarks and find that our approach achieves state-of-the-art performance on the average success rate metric among continual learning methods. Code is available at https://github.com/WilliamYue37/t-DGR.