Large Margin Few-Shot Learning

TL;DR

This paper introduces a large margin principle to enhance the generalization ability of metric-based few-shot learning methods by augmenting the loss function, leading to significant performance improvements with minimal additional computation.

Contribution

It proposes a unified framework that incorporates a large margin distance loss to improve metric learning in few-shot tasks, demonstrating effectiveness across multiple models.

Findings

Improved performance of graph neural networks and prototypical networks on few-shot tasks.

Enhanced discriminative power of the learned metric space.

Minimal computational overhead introduced by the method.

Abstract

The key issue of few-shot learning is learning to generalize. This paper proposes a large margin principle to improve the generalization capacity of metric based methods for few-shot learning. To realize it, we develop a unified framework to learn a more discriminative metric space by augmenting the classification loss function with a large margin distance loss function for training. Extensive experiments on two state-of-the-art few-shot learning methods, graph neural networks and prototypical networks, show that our method can improve the performance of existing models substantially with very little computational overhead, demonstrating the effectiveness of the large margin principle and the potential of our method.