Open Set Recognition Through Deep Neural Network Uncertainty: Does Out-of-Distribution Detection Require Generative Classifiers?

TL;DR

This paper investigates whether deep neural networks can effectively detect out-of-distribution data using uncertainty estimates, highlighting the advantages of generative models over discriminative ones for open set recognition.

Contribution

It demonstrates that combining uncertainty estimation with extreme value theory improves out-of-distribution detection, especially with generative models, challenging the necessity of generative classifiers.

Findings

Generative models outperform discriminative models in open set recognition.

Uncertainty alone is insufficient for reliable out-of-distribution detection.

Posterior-based methods enhance out-of-distribution detection performance.

Abstract

We present an analysis of predictive uncertainty based out-of-distribution detection for different approaches to estimate various models' epistemic uncertainty and contrast it with extreme value theory based open set recognition. While the former alone does not seem to be enough to overcome this challenge, we demonstrate that uncertainty goes hand in hand with the latter method. This seems to be particularly reflected in a generative model approach, where we show that posterior based open set recognition outperforms discriminative models and predictive uncertainty based outlier rejection, raising the question of whether classifiers need to be generative in order to know what they have not seen.