Improving Explainability of Softmax Classifiers Using a Prototype-Based Joint Embedding Method
This addresses the need for more interpretable and reliable confidence estimates in classification models, particularly for applications requiring trustworthy AI decisions, though it appears incremental as it builds on existing prototype and embedding techniques.
The paper tackles the problem of improving explainability in softmax classifiers by proposing a prototype-based joint embedding method that provides instance-based explanations through sampled prototypes and offers better uncertainty metrics for out-of-distribution detection compared to softmax confidence.
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.