POEM: Out-of-Distribution Detection with Posterior Sampling
This addresses the challenge of efficiently using outlier data for OOD detection in machine learning models deployed in open-world scenarios, representing a strong specific gain rather than a foundational advancement.
The paper tackles the problem of out-of-distribution detection by proposing POEM, a posterior sampling-based outlier mining framework that improves detection performance, achieving relative improvements of 42.0% and 24.2% in FPR95 on CIFAR-10 and CIFAR-100 benchmarks compared to the current best method.
Out-of-distribution (OOD) detection is indispensable for machine learning models deployed in the open world. Recently, the use of an auxiliary outlier dataset during training (also known as outlier exposure) has shown promising performance. As the sample space for potential OOD data can be prohibitively large, sampling informative outliers is essential. In this work, we propose a novel posterior sampling-based outlier mining framework, POEM, which facilitates efficient use of outlier data and promotes learning a compact decision boundary between ID and OOD data for improved detection. We show that POEM establishes state-of-the-art performance on common benchmarks. Compared to the current best method that uses a greedy sampling strategy, POEM improves the relative performance by 42.0% and 24.2% (FPR95) on CIFAR-10 and CIFAR-100, respectively. We further provide theoretical insights on the effectiveness of POEM for OOD detection.