Students are the Best Teacher: Exit-Ensemble Distillation with Multi-Exits

TL;DR

This paper introduces exit-ensemble distillation, a novel method that uses multi-exit CNN architectures to improve classification accuracy without pre-trained teachers, leveraging student-student and student-teacher collaboration.

Contribution

It presents a new knowledge distillation paradigm using multi-exit CNNs where ensemble of exits guides training, enhancing performance and convergence speed.

Findings

Significant accuracy improvements on various CNN architectures.

Faster convergence and improved training stability.

Effective without pre-trained teacher networks.

Abstract

This paper proposes a novel knowledge distillation-based learning method to improve the classification performance of convolutional neural networks (CNNs) without a pre-trained teacher network, called exit-ensemble distillation. Our method exploits the multi-exit architecture that adds auxiliary classifiers (called exits) in the middle of a conventional CNN, through which early inference results can be obtained. The idea of our method is to train the network using the ensemble of the exits as the distillation target, which greatly improves the classification performance of the overall network. Our method suggests a new paradigm of knowledge distillation; unlike the conventional notion of distillation where teachers only teach students, we show that students can also help other students and even the teacher to learn better. Experimental results demonstrate that our method achieves significant improvement of classification performance on various popular CNN architectures (VGG, ResNet, ResNeXt, WideResNet, etc.). Furthermore, the proposed method can expedite the convergence of learning with improved stability. Our code will be available on Github.