Learning to generate imaginary tasks for improving generalization in meta-learning

TL;DR

This paper introduces Adversarial Task Up-sampling (ATU), a novel method that generates imaginary tasks to enhance meta-learning generalization, outperforming existing augmentation strategies on sine regression and image classification benchmarks.

Contribution

The paper proposes a new task generation approach using a task up-sampling network trained adversarially to improve meta-learning performance.

Findings

ATU significantly improves meta-testing accuracy.

Generated tasks are of high quality and diversity.

Outperforms existing task augmentation methods.

Abstract

The success of meta-learning on existing benchmarks is predicated on the assumption that the distribution of meta-training tasks covers meta-testing tasks. Frequent violation of the assumption in applications with either insufficient tasks or a very narrow meta-training task distribution leads to memorization or learner overfitting. Recent solutions have pursued augmentation of meta-training tasks, while it is still an open question to generate both correct and sufficiently imaginary tasks. In this paper, we seek an approach that up-samples meta-training tasks from the task representation via a task up-sampling network. Besides, the resulting approach named Adversarial Task Up-sampling (ATU) suffices to generate tasks that can maximally contribute to the latest meta-learner by maximizing an adversarial loss. On few-shot sine regression and image classification datasets, we empirically validate the marked improvement of ATU over state-of-the-art task augmentation strategies in the meta-testing performance and also the quality of up-sampled tasks.