A Meta-Learning Approach for Custom Model Training

TL;DR

This paper introduces a joint transfer-meta learning approach that enhances model generalization across various few- and many-shot, few- and many-class scenarios by combining the strengths of both methods.

Contribution

It proposes a novel joint training method that integrates transfer-learning and meta-learning to improve performance on diverse target tasks.

Findings

Improved generalization in few- and many-shot scenarios.

Enhanced performance across few- and many-class tasks.

Meta-learning alone does not generalize well to many-shot, many-class cases.

Abstract

Transfer-learning and meta-learning are two effective methods to apply knowledge learned from large data sources to new tasks. In few-class, few-shot target task settings (i.e. when there are only a few classes and training examples available in the target task), meta-learning approaches that optimize for future task learning have outperformed the typical transfer approach of initializing model weights from a pre-trained starting point. But as we experimentally show, meta-learning algorithms that work well in the few-class setting do not generalize well in many-shot and many-class cases. In this paper, we propose a joint training approach that combines both transfer-learning and meta-learning. Benefiting from the advantages of each, our method obtains improved generalization performance on unseen target tasks in both few- and many-class and few- and many-shot scenarios.