The Mind's Eye: Visualizing Class-Agnostic Features of CNNs

TL;DR

This paper introduces a new method for visualizing CNN features in a class-agnostic way, enabling better understanding of what features layers respond to without modifying the original model.

Contribution

It proposes a novel visualization approach using dual-objective loss that does not require a generator network or model modifications.

Findings

Effective visualization of CNN features across different architectures.

No additional FLOPs required beyond the original network.

Enhanced interpretability of CNN layers.

Abstract

Visual interpretability of Convolutional Neural Networks (CNNs) has gained significant popularity because of the great challenges that CNN complexity imposes to understanding their inner workings. Although many techniques have been proposed to visualize class features of CNNs, most of them do not provide a correspondence between inputs and the extracted features in specific layers. This prevents the discovery of stimuli that each layer responds better to. We propose an approach to visually interpret CNN features given a set of images by creating corresponding images that depict the most informative features of a specific layer. Exploring features in this class-agnostic manner allows for a greater focus on the feature extractor of CNNs. Our method uses a dual-objective activation maximization and distance minimization loss, without requiring a generator network nor modifications to the original model. This limits the number of FLOPs to that of the original network. We demonstrate the visualization quality on widely-used architectures.