Continual Few-Shot Learning with Adversarial Class Storage

TL;DR

This paper introduces Continual Meta-Learner (CML), a novel approach combining metric-based classification, memory, and adversarial learning to enable models to learn new tasks quickly while avoiding forgetting in a continual few-shot learning setting.

Contribution

It proposes a new continual few-shot learning framework with a model-agnostic meta-learning method that effectively handles sequential tasks without catastrophic forgetting.

Findings

CML achieves state-of-the-art accuracy on MiniImageNet and CIFAR100.

CML effectively prevents catastrophic forgetting in continual learning.

CML demonstrates rapid adaptation to new tasks in few-shot scenarios.

Abstract

Humans have a remarkable ability to quickly and effectively learn new concepts in a continuous manner without forgetting old knowledge. Though deep learning has made tremendous successes on various computer vision tasks, it faces challenges for achieving such human-level intelligence. In this paper, we define a new problem called continual few-shot learning, in which tasks arrive sequentially and each task is associated with a few training samples. We propose Continual Meta-Learner (CML) to solve this problem. CML integrates metric-based classification and a memory-based mechanism along with adversarial learning into a meta-learning framework, which leads to the desirable properties: 1) it can quickly and effectively learn to handle a new task; 2) it overcomes catastrophic forgetting; 3) it is model-agnostic. We conduct extensive experiments on two image datasets, MiniImageNet and CIFAR100. Experimental results show that CML delivers state-of-the-art performance in terms of classification accuracy on few-shot learning tasks without catastrophic forgetting.