Saliency Map Based Data Augmentation

TL;DR

This paper introduces a saliency map-based data augmentation method that enhances neural network classification accuracy by restricting invariance to relevant image regions, addressing the challenge of irrelevant features.

Contribution

The paper proposes a novel saliency map-guided augmentation technique that selectively restricts invariance to relevant regions, improving classification performance.

Findings

Higher test accuracy achieved on classification tasks

Saliency-guided augmentation improves invariance control

Method effectively distinguishes relevant from irrelevant features

Abstract

Data augmentation is a commonly applied technique with two seemingly related advantages. With this method one can increase the size of the training set generating new samples and also increase the invariance of the network against the applied transformations. Unfortunately all images contain both relevant and irrelevant features for classification therefore this invariance has to be class specific. In this paper we will present a new method which uses saliency maps to restrict the invariance of neural networks to certain regions, providing higher test accuracy in classification tasks.