NSF Award
2407679
Fundamental exoplanet properties, such as mass and semimajor axis, are set by the conditions of their formation in the protoplanetary disk. They may also be determined from observations of those disks, if interpreted with appropriate tools. This project develops DECIPHER: Deep computer vision in the astrophysics of planet formation — a machine learning tool that detects and measures the mass of forming exoplanets in protoplanetary disks from observational data. It is cross-disciplinary, using cutting-edge advances in computational astrophysics, artificial intelligence (AI), and computer science. With a goal of broadening participation in STEM graduate studies, this project actively involves undergraduate students in the development of DECIPHER. Moreover, this project conducts free research-based workshops to train undergraduate students in the fundamentals of programming and data analysis.<br/><br/>DECIPHER will be released on a collaborative, open-source basis through GitHub. The AI advances leveraged are semantic image segmentation and generative adversarial networks, which will be accessed through collaborations with forefront AI researchers. The code will be developed through an integrated research-education program, where undergraduate students from the University of Georgia and other universities in the Peach State Louis Stokes Alliance for Minority Participation (PS-LSAMP) lead pivotal sections of the testing and development of DECIPHER. It has the potential to increase the speed of detection and characterization of forming exoplanets by orders of magnitude, without additional computationally expensive, ad hoc simulations, opening analysis of these disks to smaller institutions such as community colleges. Undergraduate students will help to assess DECIPHER’s ability to outperform the current state-of-the art, human-based approach to detecting exoplanets and measuring exoplanet mass from observations of their formation.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
