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

TL;DR

This paper introduces AIOL, a method for detecting both seen and unseen out-of-distribution samples using limited labeled data and abundant unlabeled data, employing adaptive selection and data augmentation for improved accuracy.

Contribution

Proposes the AIOL method that adaptively identifies ID and OOD samples from mixed unlabeled data and enhances detection of unseen OOD classes with data augmentation.

Findings

AIOL outperforms existing methods on benchmark datasets.

Adaptive temperature selection improves ID/OOD sample separation.

Data augmentation enhances unseen OOD detection accuracy.

Abstract

Out-of-Distribution (OOD) detection is essential in real-world applications, which has attracted increasing attention in recent years. However, most existing OOD detection methods require many labeled In-Distribution (ID) data, causing a heavy labeling cost. In this paper, we focus on the more realistic scenario, where limited labeled data and abundant unlabeled data are available, and these unlabeled data are mixed with ID and OOD samples. 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. Moreover, since the test data in realistic applications may contain OOD samples whose classes are not in the mixed unlabeled data (we call them unseen OOD classes), data augmentation techniques are brought into the method to further improve the performance. The experiments are conducted on various benchmark datasets, which demonstrate the superiority of our method.