Representation Learning in Anomaly Detection: Successes, Limits and a Grand Challenge

TL;DR

This paper discusses the limitations of current anomaly detection methods, emphasizing the need for new tools to address grand challenges like scientific discovery and complex image anomaly detection.

Contribution

It identifies fundamental limits of existing anomaly detection paradigms and proposes grand challenges to motivate development of advanced methods.

Findings

Limitations due to no free lunch principle in anomaly detection

Strong task priors enable better anomaly detection performance

Proposes grand challenges for future research in anomaly detection

Abstract

In this perspective paper, we argue that the dominant paradigm in anomaly detection cannot scale indefinitely and will eventually hit fundamental limits. This is due to the a no free lunch principle for anomaly detection. These limitations can be overcome when there are strong tasks priors, as is the case for many industrial tasks. When such priors do not exists, the task is much harder for anomaly detection. We pose two such tasks as grand challenges for anomaly detection: i) scientific discovery by anomaly detection ii) a "mini-grand" challenge of detecting the most anomalous image in the ImageNet dataset. We believe new anomaly detection tools and ideas would need to be developed to overcome these challenges.