Key Feature Replacement of In-Distribution Samples for Out-of-Distribution Detection

TL;DR

This paper introduces KIRBY, a method that creates surrogate out-of-distribution data by replacing key features with background features, improving OOD detection by leveraging background similarity.

Contribution

The paper proposes a novel feature replacement technique, KIRBY, that enhances OOD detection by constructing surrogate datasets through inpainting-based background feature substitution.

Findings

KIRBY outperforms state-of-the-art methods on various benchmarks.

Replacing class-discriminative features with background features improves OOD detection.

Extensive ablation studies validate each step of the KIRBY procedure.

Abstract

Out-of-distribution (OOD) detection can be used in deep learning-based applications to reject outlier samples from being unreliably classified by deep neural networks. Learning to classify between OOD and in-distribution samples is difficult because data comprising the former is extremely diverse. It has been observed that an auxiliary OOD dataset is most effective in training a "rejection" network when its samples are semantically similar to in-distribution images. We first deduce that OOD images are perceived by a deep neural network to be semantically similar to in-distribution samples when they share a common background, as deep networks are observed to incorrectly classify such images with high confidence. We then propose a simple yet effective Key In-distribution feature Replacement BY inpainting (KIRBY) procedure that constructs a surrogate OOD dataset by replacing class-discriminative features of in-distribution samples with marginal background features. The procedure can be implemented using off-the-shelf vision algorithms, where each step within the algorithm is shown to make the surrogate data increasingly similar to in-distribution data. Design choices in each step are studied extensively, and an exhaustive comparison with state-of-the-art algorithms demonstrates KIRBY's competitiveness on various benchmarks.