Explaining Convolutional Neural Networks by Tagging Filters

TL;DR

This paper introduces FilTag, a novel method for explaining CNNs by tagging filters with class-specific features, making model decisions more interpretable for non-experts and aiding error analysis.

Contribution

The paper proposes a new filter tagging approach that enhances interpretability of CNNs and assists in analyzing classification errors, especially for non-expert users.

Findings

Tags help explain individual classifications.

Tags assist in analyzing errors caused by noise.

Tags can be processed by machines for further analysis.

Abstract

Convolutional neural networks (CNNs) have achieved astonishing performance on various image classification tasks, but it is difficult for humans to understand how a classification comes about. Recent literature proposes methods to explain the classification process to humans. These focus mostly on visualizing feature maps and filter weights, which are not very intuitive for non-experts in analyzing a CNN classification. In this paper, we propose FilTag, an approach to effectively explain CNNs even to non-experts. The idea is that when images of a class frequently activate a convolutional filter, then that filter is tagged with that class. These tags provide an explanation to a reference of a class-specific feature detected by the filter. Based on the tagging, individual image classifications can then be intuitively explained in terms of the tags of the filters that the input image activates. Finally, we show that the tags are helpful in analyzing classification errors caused by noisy input images and that the tags can be further processed by machines.