Announcements of Opportunity
SURF: Announcements of Opportunity
Below are Announcements of Opportunity posted by Caltech faculty and JPL technical staff for the SURF program.
Each AO indicates whether or not it is open to non-Caltech students. If an AO is NOT open to non-Caltech students, please DO NOT contact the mentor.
Announcements of Opportunity are posted as they are received. Please check back regularly for new AO submissions! Remember: This is just one way that you can go about identifying a suitable project and/or mentor. Click here for more tips on finding a mentor.
Announcements for external summer programs are listed here.
New for 2021: Students applying for JPL projects should complete a SURF@JPL application instead of a "regular" SURF application.
Students pursuing opportunities at JPL must be
U.S. citizens or U.S. permanent residents.
|Project:||Finding Outliers in Zwicky Transient Facility data|
|Disciplines:||Data Science, Astronomy|
Lead Computational Scientist, (PMA),
|Mentor URL:||http://www.astro.caltech.edu/~aam (opens in new window)|
The Zwicky Transient facility (ZTF) is a survey to observe the northern sky (and some southern parts that can be reached) in g,r, and i filters, using a camera on the 1.2m Palomar Schmidt telescope that covers 47 sq. degrees (= 200 full moons) per image. The images are converted to catalogs of astronomical sources. ZTF DR4 released in December 2020 boasts of ~300 billion detections across well over a billion sources. A vast majority do not vary over our life-times. Many types of interesting transients and variables are being found in the remaining tens of millions of objects - hundreds of thousands of alerts are sent out in real-time per night of observing. Machine learning is being used for classification, both on the transients in real-time, and on the variables as an archival dataset. This helps reduce the amount of time spent required for spectroscopic follow-up, and concentrate on objects that push our understanding.
In order to construct priors for machine learning, good labeled datasets are required for different classes of objects. Some such datasets have been put together using known objects and some active learning for the commonest of classes (supernovae, pulsating stars, eclipsing binaries etc.). From these binary classifiers hierarchical classifiers can be put together. Objects that do not get classified into these types likely belong to more interesting types by virtue of their rarity. We will use DBSCAN to look for outliers in the data in addition to the ones caught by the hierarchical classifier. DBSCAN clusters the inputs provided into sets of objects that are like each other, and also sets aside a set of outliers. Tools/methods to be used include DBSCAN, XGBoost, active learning, and possibly additional machine learning before and after the clustering part.
Proficiency in python, jupyter notebooks, and git. Conversant with statistics basics, knowledge about linux/unix. Basic astronomy knowledge will be a plus. Knowledge about deep learning, GPUs, Mongo DB and AWS will also be a bonus.
This AO can be done under the following programs:
<< Prev Record 28 of 69 Next >> Back To List