Interpretable Image Classification with Differentiable Prototypes Assignment

TL;DR

ProtoPool is an interpretable image classification model that uses shared prototypes, enabling straightforward training without pruning, focusing on rare features, and achieving state-of-the-art accuracy with fewer prototypes.

Contribution

It introduces a fully differentiable prototype assignment and a novel focal similarity function, improving interpretability and efficiency over existing methods.

Findings

Achieves state-of-the-art accuracy on CUB-200-2011 and Stanford Cars datasets.

Reduces the number of prototypes needed for accurate classification.

Prototypes are more distinctive than those from competitive methods.

Abstract

We introduce ProtoPool, an interpretable image classification model with a pool of prototypes shared by the classes. The training is more straightforward than in the existing methods because it does not require the pruning stage. It is obtained by introducing a fully differentiable assignment of prototypes to particular classes. Moreover, we introduce a novel focal similarity function to focus the model on the rare foreground features. We show that ProtoPool obtains state-of-the-art accuracy on the CUB-200-2011 and the Stanford Cars datasets, substantially reducing the number of prototypes. We provide a theoretical analysis of the method and a user study to show that our prototypes are more distinctive than those obtained with competitive methods.