Improving Explainability of Softmax Classifiers Using a Prototype-Based Joint Embedding Method

TL;DR

This paper introduces a prototype-based joint embedding method to enhance the explainability and out-of-distribution detection capabilities of softmax classifiers by leveraging similarity sampling and latent space relationships.

Contribution

It presents a novel architecture that uses prototypes for explainability and improved OOD detection, integrating prototype sampling and relative distance learning.

Findings

Provides understandable prediction confidence through prototype sampling

Demonstrates improved out-of-distribution detection over softmax confidence

Enables instance-based explanations for model decisions

Abstract

We propose a prototype-based approach for improving explainability of softmax classifiers that provides an understandable prediction confidence, generated through stochastic sampling of prototypes, and demonstrates potential for out of distribution detection (OOD). By modifying the model architecture and training to make predictions using similarities to any set of class examples from the training dataset, we acquire the ability to sample for prototypical examples that contributed to the prediction, which provide an instance-based explanation for the model's decision. Furthermore, by learning relationships between images from the training dataset through relative distances within the model's latent space, we obtain a metric for uncertainty that is better able to detect out of distribution data than softmax confidence.