Is Fast Adaptation All You Need?

TL;DR

This paper explores alternative second-order training signals in gradient-based meta-learning, showing that minimizing interference enhances incremental learning more effectively than focusing solely on fast adaptation.

Contribution

It introduces the idea of using robustness to interference as a training signal, expanding beyond traditional fast adaptation metrics in meta-learning.

Findings

Minimizing interference leads to better incremental learning.

Representations learned with interference minimization are more robust.

Traditional fast adaptation metrics may overlook important robustness aspects.

Abstract

Gradient-based meta-learning has proven to be highly effective at learning model initializations, representations, and update rules that allow fast adaptation from a few samples. The core idea behind these approaches is to use fast adaptation and generalization -- two second-order metrics -- as training signals on a meta-training dataset. However, little attention has been given to other possible second-order metrics. In this paper, we investigate a different training signal -- robustness to catastrophic interference -- and demonstrate that representations learned by directing minimizing interference are more conducive to incremental learning than those learned by just maximizing fast adaptation.