Exploiting Mixed Unlabeled Data for Detecting Samples of Seen and Unseen Out-of-Distribution Classes

no code implementations • 13 Oct 2022 • Yi-Xuan Sun, Wei Wang

We propose the Adaptive In-Out-aware Learning (AIOL) method, in which we employ the appropriate temperature to adaptively select potential ID and OOD samples from the mixed unlabeled data and consider the entropy over them for OOD detection.

Data Augmentation Out of Distribution (OOD) Detection