Out-of-distribution Detection in Classifiers via Generation

TL;DR

This paper introduces a novel method for out-of-distribution detection in classifiers by generating effective OOD samples using a manifold learning network and training an extended classifier, improving detection performance.

Contribution

The paper proposes a new algorithm that generates comprehensive OOD samples with a manifold learning network and trains an extended classifier for better OOD detection.

Findings

The proposed method outperforms recent OOD detection approaches on MNIST and Fashion-MNIST.

Generated OOD samples effectively cover the in-distribution boundary.

The approach demonstrates consistent improvement in OOD detection accuracy.

Abstract

By design, discriminatively trained neural network classifiers produce reliable predictions only for in-distribution samples. For their real-world deployments, detecting out-of-distribution (OOD) samples is essential. Assuming OOD to be outside the closed boundary of in-distribution, typical neural classifiers do not contain the knowledge of this boundary for OOD detection during inference. There have been recent approaches to instill this knowledge in classifiers by explicitly training the classifier with OOD samples close to the in-distribution boundary. However, these generated samples fail to cover the entire in-distribution boundary effectively, thereby resulting in a sub-optimal OOD detector. In this paper, we analyze the feasibility of such approaches by investigating the complexity of producing such "effective" OOD samples. We also propose a novel algorithm to generate such samples using a manifold learning network (e.g., variational autoencoder) and then train an n+1 classifier for OOD detection, where the $n+1^{th}$ class represents the OOD samples. We compare our approach against several recent classifier-based OOD detectors on MNIST and Fashion-MNIST datasets. Overall the proposed approach consistently performs better than the others.