Continual Learning via Inter-Task Synaptic Mapping

TL;DR

This paper introduces ISYANA, a novel continual learning method that leverages inter-task and concept-to-concept relationships to improve knowledge retention and scalability in streaming task scenarios.

Contribution

ISYANA is a new approach that exploits shared information across tasks to prevent catastrophic forgetting and enhance scalability in continual learning.

Findings

ISYANA performs competitively against state-of-the-art methods.

It effectively prevents neurons from embracing conflicting concepts.

The method scales better to large problems.

Abstract

Learning from streaming tasks leads a model to catastrophically erase unique experiences it absorbs from previous episodes. While regularization techniques such as LWF, SI, EWC have proven themselves as an effective avenue to overcome this issue by constraining important parameters of old tasks from changing when accepting new concepts, these approaches do not exploit common information of each task which can be shared to existing neurons. As a result, they do not scale well to large-scale problems since the parameter importance variables quickly explode. An Inter-Task Synaptic Mapping (ISYANA) is proposed here to underpin knowledge retention for continual learning. ISYANA combines task-to-neuron relationship as well as concept-to-concept relationship such that it prevents a neuron to embrace distinct concepts while merely accepting relevant concept. Numerical study in the benchmark continual learning problems has been carried out followed by comparison against prominent continual learning algorithms. ISYANA exhibits competitive performance compared to state of the arts. Codes of ISYANA is made available in \url{https://github.com/ContinualAL/ISYANAKBS}.