Gonzalez: Exoplanet detection for direct imaging using Neural Networks / Neyrinck: The Cosmic Spiderweb

Jun 26, 2018 - 11:00 am to 12:00 pm

Campus, Varian 355

Carlos Gomez Gonzalez (Univ. of Grenoble) / Mark Neyrinck (Durham University, UK)


Direct imaging of exoplanets is a formidable challenge that has only been enabled in the last fifteen years, thanks to advances in instrumentation and observing techniques for high-contrast imaging (HCI). The development of innovative image processing techniques will be key for pushing the detection limits of current and future HCI instruments and surveys. After a quick overview of the existing image post-processing techniques for exoplanet direct imaging, I will present my work integrating data science best practices into the field of HCI. These efforts can be summarized in three areas: the development of open and reusable scientific software [1], the reformulation of the exoplanet direct detection task as a supervised learning problem [2], and the organization of a community-wide data challenge on exoplanet direct imaging [3].


[1] https://github.com/vortex-exoplanet/VIP

[2] https://www.aanda.org/articles/aa/abs/2018/05/aa31961-17/aa31961-17.html

[3] https://carlgogo.github.io/exoimaging_challenge/