What do we learn? Debunking the Myth of Unsupervised Outlier Detection

TL;DR

This paper critically examines the capabilities of auto-encoders in out-of-distribution detection and introduces deformable auto-encoders that outperform existing methods in identifying anomalies and pathologies.

Contribution

It challenges assumptions about auto-encoders' ability to detect outliers and proposes MorphAEus, a novel deformable auto-encoder architecture with improved anomaly detection performance.

Findings

State-of-the-art auto-encoders often fail to constrain the latent space or accurately reconstruct abnormal patterns.

Proposed MorphAEus learns perceptually aware priors and adapts morphometry, improving anomaly detection.

MorphAEus outperforms existing unsupervised methods in detecting out-of-distribution samples and pathologies.

Abstract

Even though auto-encoders (AEs) have the desirable property of learning compact representations without labels and have been widely applied to out-of-distribution (OoD) detection, they are generally still poorly understood and are used incorrectly in detecting outliers where the normal and abnormal distributions are strongly overlapping. In general, the learned manifold is assumed to contain key information that is only important for describing samples within the training distribution, and that the reconstruction of outliers leads to high residual errors. However, recent work suggests that AEs are likely to be even better at reconstructing some types of OoD samples. In this work, we challenge this assumption and investigate what auto-encoders actually learn when they are posed to solve two different tasks. First, we propose two metrics based on the Fr\'echet inception distance (FID) and confidence scores of a trained classifier to assess whether AEs can learn the training distribution and reliably recognize samples from other domains. Second, we investigate whether AEs are able to synthesize normal images from samples with abnormal regions, on a more challenging lung pathology detection task. We have found that state-of-the-art (SOTA) AEs are either unable to constrain the latent manifold and allow reconstruction of abnormal patterns, or they are failing to accurately restore the inputs from their latent distribution, resulting in blurred or misaligned reconstructions. We propose novel deformable auto-encoders (MorphAEus) to learn perceptually aware global image priors and locally adapt their morphometry based on estimated dense deformation fields. We demonstrate superior performance over unsupervised methods in detecting OoD and pathology.