Mixture Outlier Exposure: Towards Out-of-Distribution Detection in Fine-grained Environments

TL;DR

This paper introduces MixOE, a novel training method that improves out-of-distribution detection in fine-grained environments by mixing inlier data with outliers to better cover the spectrum of potential OOD samples.

Contribution

The paper presents MixOE, a new approach that enhances OOD detection in fine-grained scenarios by mixing inlier and outlier data during training, addressing coverage gaps of existing methods.

Findings

MixOE outperforms existing methods in fine-grained OOD detection.

Constructed large-scale fine-grained test environments for evaluation.

Prediction confidence linearly decays from ID to OOD with MixOE.

Abstract

Many real-world scenarios in which DNN-based recognition systems are deployed have inherently fine-grained attributes (e.g., bird-species recognition, medical image classification). In addition to achieving reliable accuracy, a critical subtask for these models is to detect Out-of-distribution (OOD) inputs. Given the nature of the deployment environment, one may expect such OOD inputs to also be fine-grained w.r.t. the known classes (e.g., a novel bird species), which are thus extremely difficult to identify. Unfortunately, OOD detection in fine-grained scenarios remains largely underexplored. In this work, we aim to fill this gap by first carefully constructing four large-scale fine-grained test environments, in which existing methods are shown to have difficulties. Particularly, we find that even explicitly incorporating a diverse set of auxiliary outlier data during training does not provide sufficient coverage over the broad region where fine-grained OOD samples locate. We then propose Mixture Outlier Exposure (MixOE), which mixes ID data and training outliers to expand the coverage of different OOD granularities, and trains the model such that the prediction confidence linearly decays as the input transitions from ID to OOD. Extensive experiments and analyses demonstrate the effectiveness of MixOE for building up OOD detector in fine-grained environments. The code is available at https://github.com/zjysteven/MixOE.