I'm an Research Scientist at the Space Science Institute's
Center for Extrasolar planetary Systems.
My main research interest is the development of machine learning applications to better extract information from astronomical data sets, especially in the puruit of exoplanet characterization. Having recently workd for the Space Telescope Science Institute as the Exoplanet and Machine Learning Support Scientist on the James Webb Space Telescope Mission, I developed anomaly detection algorithms for both spatially and temporally correlated anomaly signals. I strongly enjoy collaborating with scientists, engineers, and adminstrators on developing the best platform for exoplanet characterization from space telescopes.
Moreover, I was the Deputy Project Manager for the Exoplanet Characterization Toolkit, and am still proud to collaborate with them to develop our "Proposal-to-Publication" Exoplanetary Science Engine
My full list of publications is available
but here are a few recent highlights:
Fraine, J.D., Wakeford, H.R., Kataria, T., Stevenson, K.B., OST Exoplanet Characterization Team, 2018
"Transiting Exoplanet Characterization Beyond 2030: A Case for Observing Giant Planets with Giant Telescopes"
National Academy of Sciences Exoplanet Science Strategy
Fraine J.D., Deming, D., Benneke, B., Knutson, H., Jordán, A., Espinoza, N., Madhusudhan, N., Wilkins, A., Todorov, K., 2014:
"Water Vapour Absorption from the Clear Atmosphere of an Exo-Neptune", Nature, Vol. 513, Issue 7519, pp. 526-529.
Fraine J.D., Deming D., Gillon M., Jehin E., Demory B.O., Benneke B., Seager S., Lewis N.K., Knutson H., and Désert J.M., 2013:
"Spitzer Transits of the Super-Earth GJ 1214 b and Implications for Its Atmosphere", ApJ, Vol. 765, Issue 2, article id. 127.
The vast majority of my day-to-day work life is to develop python code for scientific and engineering data analysis. Almost all of it can be found at my Exowanderer Github page.
I am particularly proud of these projects:
Teaching a computer to do tricks can be hard; teaching it to do physics can be harder. My student and I are currently developing a physics model emulator that can rapidly MIMIC the physical behaviour of stellar and exoplanetary atmospheric signals by using genetic algorithms with generative deep learning to rapidly develop new physics based models from large databases of well-vetted models (e.g. PHOENIX, ATLAS).
Spitzer Machine Learning Calibration
Dr. Jessica Krick (IPAC), Dr. James Ingalls (IPAC), and I (SSI) are developing several machine learning pathways to use existing Spitzer Space Telescope time-series calibration data to pre-calibrate all exoplanet time-series observations with the Sptizer Space Telescope. Our foci are eXtreme Gradient Boosted Trees, Gradient Adpated Gaussian Kernel Regression, and Deep Convolutional Neural Networks, respectively.
Time series photometric pipeline optimized for the Spitzer Space Telescope, with hooks for the James Webb Space Telescope and various ground based telescopes, such at the Kitt Peak National Observatory. See the paper.
When you load the code, it kindly reminds you that "Not all who wander are lost."
A self-calibration modeling code that simultaneously models the Sptizer Space Telescope's intrapixel noise effect using the major three models (BLISS, KRData, and PLD) to extract and compare the exoplanetary atmospheric physics buried inside the 100x greater correlated noise signal from the telescope; all of which in a pure Bayesian framework.