Meta-Baseline: Exploring Simple Meta-Learning for Few-Shot Learning

TL;DR

This paper investigates a simple meta-learning approach applied to a pre-trained model's evaluation metric, demonstrating competitive results in few-shot learning and providing insights into the trade-offs with traditional classification methods.

Contribution

It introduces a straightforward meta-learning method over a pre-trained model for few-shot learning, challenging the necessity of complex meta-learning algorithms.

Findings

Achieves competitive performance on standard benchmarks

Provides analysis of trade-offs between meta-learning and whole-classification

Highlights the effectiveness of simple meta-learning approaches

Abstract

Meta-learning has been the most common framework for few-shot learning in recent years. It learns the model from collections of few-shot classification tasks, which is believed to have a key advantage of making the training objective consistent with the testing objective. However, some recent works report that by training for whole-classification, i.e. classification on the whole label-set, it can get comparable or even better embedding than many meta-learning algorithms. The edge between these two lines of works has yet been underexplored, and the effectiveness of meta-learning in few-shot learning remains unclear. In this paper, we explore a simple process: meta-learning over a whole-classification pre-trained model on its evaluation metric. We observe this simple method achieves competitive performance to state-of-the-art methods on standard benchmarks. Our further analysis shed some light on understanding the trade-offs between the meta-learning objective and the whole-classification objective in few-shot learning.