Continuous Learning in Single-Incremental-Task Scenarios

TL;DR

This paper addresses the challenge of continual learning in single-incremental-task scenarios, proposing a new method called AR1 that improves performance with minimal overhead.

Contribution

The paper introduces AR1, a novel approach combining architectural and regularization strategies tailored for single-incremental-task learning.

Findings

AR1 outperforms existing strategies on CORe50 and iCIFAR-100 datasets.

AR1 has very low memory and computation overhead.

It is suitable for online learning scenarios.

Abstract

It was recently shown that architectural, regularization and rehearsal strategies can be used to train deep models sequentially on a number of disjoint tasks without forgetting previously acquired knowledge. However, these strategies are still unsatisfactory if the tasks are not disjoint but constitute a single incremental task (e.g., class-incremental learning). In this paper we point out the differences between multi-task and single-incremental-task scenarios and show that well-known approaches such as LWF, EWC and SI are not ideal for incremental task scenarios. A new approach, denoted as AR1, combining architectural and regularization strategies is then specifically proposed. AR1 overhead (in term of memory and computation) is very small thus making it suitable for online learning. When tested on CORe50 and iCIFAR-100, AR1 outperformed existing regularization strategies by a good margin.