Clustered Task-Aware Meta-Learning by Learning From Learning Paths

To enable effective learning of new tasks with only a few examples, meta-learning acquires common knowledge from the existing tasks with a globally shared meta-learner. To further address the problem of task heterogeneity, recent developments balance between customization and generalization by incorporating task clustering to generate task-aware modulation to be applied to the global meta-learner. However, these methods learn task representation mostly from the features ofinput data, while the task-specific optimization process with respect to the base-learner is often neglected. In this work, we propose a Clustered Task-Aware Meta-Learning (CTML) framework with task representation learned from both features and learning paths. We first conduct rehearsed task learning from the common initialization, and collect a set of geometric quantities that adequately describes this learning path. By inputting this set of values into a meta path learner, we automatically abstract path representation optimized for downstream clustering and modulation. Aggregating the path and feature representations results in an improved task representation. To further improve inference efficiency, we devise a shortcut tunnel to bypass the rehearsed learning process at a meta-testing time. Extensive experiments on two real-world application domains: few-shot image classification and cold-start recommendation demonstrate the superiority of CTML compared to state-of-the-art methods. We provide our code at https://github.com/didiya0825.

I. Introduction

The astonishing performance of deep learning relies on large amounts of data, which are not always available. Humans, on the other hand, are able to learn new tasks much more quickly, leveraging prior experience to relate knowledge among tasks. Inspired by this property of human intelligence, meta-learning (also known as learning to learn) [1] acquires transferable knowledge from existing tasks in the form of embedding functions [2], [3], [4], initial parameters [5], [6], optimization strategies [7], [8], or models that directly map training samples to network parameters [9], [10]. Recent developments adopt more advanced techniques like transductive inference [11], [12] and causal intervention [13] to achieve further improvements. Although meta-learning has shown success in fields like few-shot image classification and cold-start recommendation, most of them typically assume that all the tasks are drawn from a single distribution and face the challenge of handling tasks that come from different underlying distributions, a problem known as task heterogeneity [14], [15], [16].