Approximations to the Fisher Information Metric of Deep Generative Models for Out-Of-Distribution Detection

no code implementations • 3 Mar 2024 • Sam Dauncey, Chris Holmes, Christopher Williams, Fabian Falck

In this work, we analyse using the gradient of a data point with respect to the parameters of the deep generative model for OOD detection, based on the simple intuition that OOD data should have larger gradient norms than training data.

Out-of-Distribution Detection Out of Distribution (OOD) Detection