Task-Aware Information Routing from Common Representation Space in Lifelong Learning

TL;DR

This paper introduces TAMiL, a task-aware continual learning method inspired by brain processes, which uses attention modules and autoencoders to selectively route task-specific information, reducing interference and catastrophic forgetting.

Contribution

The paper proposes TAMiL, a novel continual learning framework that employs task-attention modules and autoencoders to improve knowledge transfer and mitigate forgetting.

Findings

Outperforms state-of-the-art rehearsal-based and dynamic sparse methods.

Effectively reduces catastrophic forgetting and task-recency bias.

Bridges gap between fixed capacity and parameter isolation approaches.

Abstract

Intelligent systems deployed in the real world suffer from catastrophic forgetting when exposed to a sequence of tasks. Humans, on the other hand, acquire, consolidate, and transfer knowledge between tasks that rarely interfere with the consolidated knowledge. Accompanied by self-regulated neurogenesis, continual learning in the brain is governed by a rich set of neurophysiological processes that harbor different types of knowledge, which are then integrated by conscious processing. Thus, inspired by the Global Workspace Theory of conscious information access in the brain, we propose TAMiL, a continual learning method that entails task-attention modules to capture task-specific information from the common representation space. We employ simple, undercomplete autoencoders to create a communication bottleneck between the common representation space and the global workspace, allowing only the task-relevant information to the global workspace, thus greatly reducing task interference. Experimental results show that our method outperforms state-of-the-art rehearsal-based and dynamic sparse approaches and bridges the gap between fixed capacity and parameter isolation approaches while being scalable. We also show that our method effectively mitigates catastrophic forgetting while being well-calibrated with reduced task-recency bias.