SmoothGrad: removing noise by adding noise

TL;DR

SmoothGrad is a technique that enhances gradient-based sensitivity maps for deep network explanations by adding noise and averaging, resulting in clearer visualizations of influential pixels.

Contribution

The paper introduces SmoothGrad, a simple yet effective method to improve the clarity of gradient-based explanations for deep networks.

Findings

SmoothGrad produces sharper sensitivity maps.

Adding noise and averaging enhances interpretability.

The method is easy to implement and improves visualization quality.

Abstract

Explaining the output of a deep network remains a challenge. In the case of an image classifier, one type of explanation is to identify pixels that strongly influence the final decision. A starting point for this strategy is the gradient of the class score function with respect to the input image. This gradient can be interpreted as a sensitivity map, and there are several techniques that elaborate on this basic idea. This paper makes two contributions: it introduces SmoothGrad, a simple method that can help visually sharpen gradient-based sensitivity maps, and it discusses lessons in the visualization of these maps. We publish the code for our experiments and a website with our results.