Speakers
Description
The Zwicky Transient Facility (ZTF) is an optical survey telescope that observes the northern sky every night. Lightcurves in $g$,$r$, and $i$ have been obtained of than a billion stars down to magnitude 20.5. Identification and classification of all variables in this huge dataset is required for multiple science cases. ZTF's volume of data, multicolour lightcurves, and (highly) irregular cadence contains a lot of information. These attributes also make classification a challenging problem, but also an excellent testing ground to learn how to deal with these challenges.
I will present our efforts to classify ZTF variable stars, how we dealt with these challenges and the design choices we made in the process. This includes the technical challenge of processing billions of lightcurves, the design of a comprehensive classification scheme and use of many one-vs-all classifiers, and our active learning approach to continuously correct and improve the classifiers.
| Main Topic | Classification and regression |
|---|---|
| Secondary Topic | Active Learning |
| Participation mode | In person |
