Self-supervised Knowledge Distillation for Few-shot Learning

TL;DR

This paper introduces a two-stage self-supervised knowledge distillation method to enhance feature representations for few-shot learning, outperforming existing approaches through entropy maximization and distillation.

Contribution

It proposes a novel two-stage training process combining entropy maximization and student-teacher distillation to improve few-shot learning performance.

Findings

Self-supervised pretraining outperforms current state-of-the-art methods.

Two-stage process yields significant improvements in few-shot tasks.

Code availability facilitates reproducibility and further research.

Abstract

Real-world contains an overwhelmingly large number of object classes, learning all of which at once is infeasible. Few shot learning is a promising learning paradigm due to its ability to learn out of order distributions quickly with only a few samples. Recent works [7, 41] show that simply learning a good feature embedding can outperform more sophisticated meta-learning and metric learning algorithms for few-shot learning. In this paper, we propose a simple approach to improve the representation capacity of deep neural networks for few-shot learning tasks. We follow a two-stage learning process: First, we train a neural network to maximize the entropy of the feature embedding, thus creating an optimal output manifold using a self-supervised auxiliary loss. In the second stage, we minimize the entropy on feature embedding by bringing self-supervised twins together, while constraining the manifold with student-teacher distillation. Our experiments show that, even in the first stage, self-supervision can outperform current state-of-the-art methods, with further gains achieved by our second stage distillation process. Our codes are available at: https://github.com/brjathu/SKD.