Paying More Attention to Attention: Improving the Performance of Convolutional Neural Networks via Attention Transfer

TL;DR

This paper introduces attention transfer techniques for CNNs, where a student network learns from a teacher's attention maps, leading to significant performance improvements across multiple datasets and architectures.

Contribution

It proposes novel methods for attention transfer in CNNs, demonstrating how mimicking attention maps enhances model performance.

Findings

Attention transfer improves CNN accuracy.

Methods are effective across various datasets.

Code and models are publicly available.

Abstract

Attention plays a critical role in human visual experience. Furthermore, it has recently been demonstrated that attention can also play an important role in the context of applying artificial neural networks to a variety of tasks from fields such as computer vision and NLP. In this work we show that, by properly defining attention for convolutional neural networks, we can actually use this type of information in order to significantly improve the performance of a student CNN network by forcing it to mimic the attention maps of a powerful teacher network. To that end, we propose several novel methods of transferring attention, showing consistent improvement across a variety of datasets and convolutional neural network architectures. Code and models for our experiments are available at https://github.com/szagoruyko/attention-transfer