A Demonstration of AutoOD: A Self-tuning Anomaly Detection System

Download PDF

Abstract

Anomaly detection is a critical task in applications like preventing financial fraud, system malfunctions, and cybersecurity attacks. While previously developed research has offered a plethora of anomaly detection algorithms, effective anomaly detection remains challenging for users due to the manual tuning process where model developers must determine which of these many algorithms is best suited to their particular domain and tune many parameters by hand to make the chosen algorithm perform well. This demonstration showcases STAND, the first self- tuning anomaly detection system which frees users from this tedious, manual tuning process. Its human-in-the-loop components effectively leverage domain user feedback to meet application specific needs. STAND outperforms the best unsupervised anomaly detection methods with a performance similar to supervised anomaly classification models which utilize ground truth labels. In this demonstration, we design an easy-to-use interface where users can gain insights on STAND’s self-tuning process and use their understanding on the dataset to steer the anomaly detection process through feedback.