Meta-Learning without Memorization

TL;DR

This paper introduces a meta-regularization method based on information theory that enables meta-learning models to adapt effectively without requiring mutually-exclusive tasks, broadening applicability and improving performance.

Contribution

The authors propose a novel meta-regularization objective that allows meta-learning from non-mutually-exclusive tasks, overcoming a key limitation of prior meta-learning algorithms.

Findings

Outperforms standard meta-learning in practical, non-mutually-exclusive task settings

Applicable to both contextual and gradient-based meta-learning algorithms

Enables effective learning in domains where traditional meta-learning fails

Abstract

The ability to learn new concepts with small amounts of data is a critical aspect of intelligence that has proven challenging for deep learning methods. Meta-learning has emerged as a promising technique for leveraging data from previous tasks to enable efficient learning of new tasks. However, most meta-learning algorithms implicitly require that the meta-training tasks be mutually-exclusive, such that no single model can solve all of the tasks at once. For example, when creating tasks for few-shot image classification, prior work uses a per-task random assignment of image classes to N-way classification labels. If this is not done, the meta-learner can ignore the task training data and learn a single model that performs all of the meta-training tasks zero-shot, but does not adapt effectively to new image classes. This requirement means that the user must take great care in designing the tasks, for example by shuffling labels or removing task identifying information from the inputs. In some domains, this makes meta-learning entirely inapplicable. In this paper, we address this challenge by designing a meta-regularization objective using information theory that places precedence on data-driven adaptation. This causes the meta-learner to decide what must be learned from the task training data and what should be inferred from the task testing input. By doing so, our algorithm can successfully use data from non-mutually-exclusive tasks to efficiently adapt to novel tasks. We demonstrate its applicability to both contextual and gradient-based meta-learning algorithms, and apply it in practical settings where applying standard meta-learning has been difficult. Our approach substantially outperforms standard meta-learning algorithms in these settings.