Continual Unsupervised Representation Learning

TL;DR

This paper introduces CURL, a novel method for unsupervised continual learning that dynamically infers tasks, expands its capacity, and mitigates forgetting, demonstrated on MNIST and Omniglot without relying on task labels.

Contribution

CURL is the first approach to address unsupervised continual learning with task inference, dynamic expansion, and rehearsal, applicable to both unsupervised and supervised settings.

Findings

Effective in unsupervised learning on MNIST and Omniglot

Outperforms prior methods in i.i.d. and incremental class learning

Handles abrupt, smooth, and shuffled task transitions

Abstract

Continual learning aims to improve the ability of modern learning systems to deal with non-stationary distributions, typically by attempting to learn a series of tasks sequentially. Prior art in the field has largely considered supervised or reinforcement learning tasks, and often assumes full knowledge of task labels and boundaries. In this work, we propose an approach (CURL) to tackle a more general problem that we will refer to as unsupervised continual learning. The focus is on learning representations without any knowledge about task identity, and we explore scenarios when there are abrupt changes between tasks, smooth transitions from one task to another, or even when the data is shuffled. The proposed approach performs task inference directly within the model, is able to dynamically expand to capture new concepts over its lifetime, and incorporates additional rehearsal-based techniques to deal with catastrophic forgetting. We demonstrate the efficacy of CURL in an unsupervised learning setting with MNIST and Omniglot, where the lack of labels ensures no information is leaked about the task. Further, we demonstrate strong performance compared to prior art in an i.i.d setting, or when adapting the technique to supervised tasks such as incremental class learning.