Real-bogus scores for active anomaly detection

TL;DR

This paper demonstrates that incorporating machine learning-based real-bogus scores into active anomaly detection pipelines in astronomical surveys significantly reduces artifacts, enhancing the identification of genuine astrophysical objects.

Contribution

The study introduces the use of real-bogus scores from machine learning classifiers into active anomaly detection, improving artifact rejection in large-scale astronomical data analysis.

Findings

Real-bogus classifiers achieve ROC-AUC scores of 0.93-0.95.

Inclusion of real-bogus scores reduces artifacts from 27 to 3 in the detection pipeline.

Active anomaly detection with real-bogus scores increases the yield of interesting astrophysical objects.

Abstract

In the task of anomaly detection in modern time-domain photometric surveys, the primary goal is to identify astrophysically interesting, rare, and unusual objects among a large volume of data. Unfortunately, artifacts -- such as plane or satellite tracks, bad columns on CCDs, and ghosts -- often constitute significant contaminants in results from anomaly detection analysis. In such contexts, the Active Anomaly Discovery (AAD) algorithm allows tailoring the output of anomaly detection pipelines according to what the expert judges to be scientifically interesting. We demonstrate how the introduction real-bogus scores, obtained from a machine learning classifier, improves the results from AAD. Using labeled data from the SNAD ZTF knowledge database, we train four real-bogus classifiers: XGBoost, CatBoost, Random Forest, and Extremely Randomized Trees. All the models perform real-bogus classification with similar effectiveness, achieving ROC-AUC scores ranging from 0.93 to 0.95. Consequently, we select the Random Forest model as the main model due to its simplicity and interpretability. The Random Forest classifier is applied to 67 million light curves from ZTF DR17. The output real-bogus score is used as an additional feature for two anomaly detection algorithms: static Isolation Forest and AAD. While results from Isolation Forest remained unchanged, the number of artifacts detected by the active approach decreases significantly with the inclusion of the real-bogus score, from 27 to 3 out of 100. We conclude that incorporating the real-bogus classifier result as an additional feature in the active anomaly detection pipeline significantly reduces the number of artifacts in the outputs, thereby increasing the incidence of astrophysically interesting objects presented to human experts.