Taxonomic Class Incremental Learning

TL;DR

This paper introduces Taxonomic Class Incremental Learning (TCIL), a new continual learning setup based on taxonomic class hierarchies, and demonstrates its effectiveness with improved accuracy on CIFAR-100 and ImageNet-100.

Contribution

It proposes the TCIL problem setting, unifies existing CIL approaches with taxonomic learning, and introduces a new parameter inheritance scheme for knowledge transfer.

Findings

TCIL outperforms state-of-the-art methods by 2% on CIFAR-100

TCIL outperforms state-of-the-art methods by 3% on ImageNet-100

Taxonomic curricula better mimic human learning processes

Abstract

The problem of continual learning has attracted rising attention in recent years. However, few works have questioned the commonly used learning setup, based on a task curriculum of random class. This differs significantly from human continual learning, which is guided by taxonomic curricula. In this work, we propose the Taxonomic Class Incremental Learning (TCIL) problem. In TCIL, the task sequence is organized based on a taxonomic class tree. We unify existing approaches to CIL and taxonomic learning as parameter inheritance schemes and introduce a new such scheme for the TCIL learning. This enables the incremental transfer of knowledge from ancestor to descendant class of a class taxonomy through parameter inheritance. Experiments on CIFAR-100 and ImageNet-100 show the effectiveness of the proposed TCIL method, which outperforms existing SOTA methods by 2% in terms of final accuracy on CIFAR-100 and 3% on ImageNet-100.