Searching for different AGN populations in massive datasets with Machine Learning

Brightness variations of active galactic nuclei (AGNs) offer key insights into their physical emission mechanisms and related phenomena. These variations also provide us an alternative way to identify AGN candidates that could be missed by more traditional selection techniques. In this talk I will first introduce the ALeRCE light curve classifier, that uses a hierarchical imbalanced Random Forest and variability features computed from ZTF light curves, to classify each source into more than 15 subclasses, including three classes of AGNs (host-dominated, core-dominated, and Blazar). Then, I will present  an anomaly detection (AD) technique designed to identify AGN light curves with anomalous behaviours in massive datasets, like the ZTF data releases. The main aim of this technique is to identify changing-state AGNs (CSAGNs) at different stages of the transition, but it can also be used for more general purposes, such as cleaning massive datasets for AGN variability analyses.