Negative Margin Matters: Understanding Margin in Few-shot Classification

TL;DR

This paper proposes a negative margin loss for metric learning in few-shot classification, which outperforms traditional methods by improving the discrimination of novel classes through a counterintuitive approach.

Contribution

It introduces a negative margin loss for few-shot learning and provides empirical and theoretical analysis explaining its effectiveness for novel class discrimination.

Findings

Negative margin loss outperforms softmax loss on benchmarks.

Negative margin reduces discriminability for training classes but benefits novel class separation.

Theoretical analysis supports the empirical benefits of negative margin in few-shot learning.

Abstract

This paper introduces a negative margin loss to metric learning based few-shot learning methods. The negative margin loss significantly outperforms regular softmax loss, and achieves state-of-the-art accuracy on three standard few-shot classification benchmarks with few bells and whistles. These results are contrary to the common practice in the metric learning field, that the margin is zero or positive. To understand why the negative margin loss performs well for the few-shot classification, we analyze the discriminability of learned features w.r.t different margins for training and novel classes, both empirically and theoretically. We find that although negative margin reduces the feature discriminability for training classes, it may also avoid falsely mapping samples of the same novel class to multiple peaks or clusters, and thus benefit the discrimination of novel classes. Code is available at https://github.com/bl0/negative-margin.few-shot.