This Looks Better than That: Better Interpretable Models with ProtoPNeXt

TL;DR

ProtoPNeXt is a new framework that enhances prototypical-part models, achieving state-of-the-art accuracy and improved interpretability through Bayesian hyperparameter tuning and angular prototype similarity.

Contribution

The paper introduces ProtoPNeXt, a framework that improves training stability, accuracy, and interpretability of prototypical-part models for computer vision tasks.

Findings

Achieved new state-of-the-art accuracy on CUB-200 dataset.

Enhanced model interpretability with minimal accuracy trade-offs.

Demonstrated the effectiveness of Bayesian hyperparameter tuning and angular similarity metrics.

Abstract

Prototypical-part models are a popular interpretable alternative to black-box deep learning models for computer vision. However, they are difficult to train, with high sensitivity to hyperparameter tuning, inhibiting their application to new datasets and our understanding of which methods truly improve their performance. To facilitate the careful study of prototypical-part networks (ProtoPNets), we create a new framework for integrating components of prototypical-part models -- ProtoPNeXt. Using ProtoPNeXt, we show that applying Bayesian hyperparameter tuning and an angular prototype similarity metric to the original ProtoPNet is sufficient to produce new state-of-the-art accuracy for prototypical-part models on CUB-200 across multiple backbones. We further deploy this framework to jointly optimize for accuracy and prototype interpretability as measured by metrics included in ProtoPNeXt. Using the same resources, this produces models with substantially superior semantics and changes in accuracy between +1.3% and -1.5%. The code and trained models will be made publicly available upon publication.