Leveraging Activations for Superpixel Explanations

TL;DR

This paper introduces Neuro-Activated Superpixels (NAS), a novel method that extracts meaningful image segments directly from neural network activations, improving interpretability and weakly supervised object localization without additional training.

Contribution

The paper presents NAS, a new segmentation approach from activations that enhances saliency interpretation and object localization in deep neural networks without fine-tuning.

Findings

NAS improves weakly supervised object localization performance.

Aggregation of NAS with existing saliency methods clarifies their interpretations.

Reveals limitations of the area under the relevance curve metric.

Abstract

Saliency methods have become standard in the explanation toolkit of deep neural networks. Recent developments specific to image classifiers have investigated region-based explanations with either new methods or by adapting well-established ones using ad-hoc superpixel algorithms. In this paper, we aim to avoid relying on these segmenters by extracting a segmentation from the activations of a deep neural network image classifier without fine-tuning the network. Our so-called Neuro-Activated Superpixels (NAS) can isolate the regions of interest in the input relevant to the model's prediction, which boosts high-threshold weakly supervised object localization performance. This property enables the semi-supervised semantic evaluation of saliency methods. The aggregation of NAS with existing saliency methods eases their interpretation and reveals the inconsistencies of the widely used area under the relevance curve metric.