Deep Mutual Learning

TL;DR

Deep Mutual Learning (DML) enables multiple student networks to learn collaboratively and improve performance without relying on a pre-trained teacher, outperforming traditional distillation methods on recognition benchmarks.

Contribution

This paper introduces a novel mutual learning strategy where multiple students learn collaboratively, eliminating the need for a pre-trained teacher and enhancing performance.

Findings

Mutual learning improves accuracy across various architectures.

DML outperforms traditional teacher-student distillation.

Simple student networks can achieve state-of-the-art results.

Abstract

Model distillation is an effective and widely used technique to transfer knowledge from a teacher to a student network. The typical application is to transfer from a powerful large network or ensemble to a small network, that is better suited to low-memory or fast execution requirements. In this paper, we present a deep mutual learning (DML) strategy where, rather than one way transfer between a static pre-defined teacher and a student, an ensemble of students learn collaboratively and teach each other throughout the training process. Our experiments show that a variety of network architectures benefit from mutual learning and achieve compelling results on CIFAR-100 recognition and Market-1501 person re-identification benchmarks. Surprisingly, it is revealed that no prior powerful teacher network is necessary -- mutual learning of a collection of simple student networks works, and moreover outperforms distillation from a more powerful yet static teacher.