Learning Robust Convolutional Neural Networks with Relevant Feature Focusing via Explanations

TL;DR

This paper introduces ReFF, a regularization method using explanations to focus CNNs on task-relevant features, improving robustness against distribution shifts without extra inference costs.

Contribution

ReFF is a novel regularization approach that leverages explanation outputs to enhance CNN focus on relevant features, applicable to existing models without additional inference overhead.

Findings

ReFF improves CNN accuracy under distribution shifts.

CNNs trained with ReFF focus more on task-relevant features.

ReFF does not increase inference cost at test time.

Abstract

Existing image recognition techniques based on convolutional neural networks (CNNs) basically assume that the training and test datasets are sampled from i.i.d distributions. However, this assumption is easily broken in the real world because of the distribution shift that occurs when the co-occurrence relations between objects and backgrounds in input images change. Under this type of distribution shift, CNNs learn to focus on features that are not task-relevant, such as backgrounds from the training data, and degrade their accuracy on the test data. To tackle this problem, we propose relevant feature focusing (ReFF). ReFF detects task-relevant features and regularizes CNNs via explanation outputs (e.g., Grad-CAM). Since ReFF is composed of post-hoc explanation modules, it can be easily applied to off-the-shelf CNNs. Furthermore, ReFF requires no additional inference cost at test time because it is only used for regularization while training. We demonstrate that CNNs trained with ReFF focus on features relevant to the target task and that ReFF improves the test-time accuracy.