Anomaly Detection in the Zwicky Transient Facility
Hunting for anomalous Astronomical Transients using Machine Learning.
(Work in progress: to be further updated)
The Zwicky Transient Facility (ZTF) is a fully automated, wide-field astronomical survey dedicated to the search for transient astronomical events, such as supernovae, black hole mergers, and other variables. Our project aims to build upon the classification pipeline of ZTF, which uses machine-learning methods to classify light curves dynamically. In particular, this project aims to refine the existing anomaly detection pipeline and apply it to billions of astronomical sources.
Anomalous transients are objects that do not fit into any of the known classes of transients. The presence of anomalies in a dataset such as ours suggests a lack of our knowledge or an indication that we may be searching in the incorrect parameter spaces.