Planetary Science Directorate

SOUTHWEST RESEARCH INSTITUTE, BOULDER OFFICE

Upcoming SwRI Boulder Colloquia

Colloquia are normally on Tuesdays at 11:00 am in the 4th-floor conference room, except as indicated below in bold text.
Show previous colloquia

For questions or suggestions for speakers, please contact the SwRI colloquium organizers:
Hannah Kaplan, 720-208-7208 or kaplan(at)boulder.swri.edu
Derek Lamb, 720-208-7207 or derek(at)boulder.swri.edu
Katie Primm, 720-240-0124 or kprimm(at)boulder.swri.edu
Raluca Rufu, 303-226-0879 or raluca(at)boulder.swri.edu
Julien Salmon, 720-208-7203 or julien(at)boulder.swri.edu
Kelsi Singer, 303-226-5910 or ksinger(at)boulder.swri.edu

To be added to the SwRI Boulder Colloquia email list, please contact Kelsi Singer, ksinger(at)boulder.swri.edu

Tue Nov 20, 201811:00 am Saverio Cambioni University of Arizona Application of machine learning to planetary science
Abstract: Machine Learning (ML) is a subfield of data analysis that lies at the cornerstone between statistical methods and computer science, as well as at the core of artificial intelligence. State-of-the-art ML techniques allow for several advantages: they can streamline the generation of data sets to most efficiently explore regions of interest in a large parameter space; and they can perform accurate mappings of initial conditions and end states, with associated probabilities, taking into account a high-dimensional parameter space. I will discuss the potentialities of coupling ML to planetary science by reporting on our current work on giant impact studies (Cambioni et al., 2018, EPSC) and characterization of thermal properties of asteroids (Cambioni et al., submitted). Giant impacts heavily influenced the final configuration and geochemistry of the terrestrial planets; we use ML to explore a rich dataset of giant impact simulations in a supervised fashion. This new methodology produces mappings of giant impact outcomes in an N-Dimensional (N-D) parameter space, e.g., mass of target, target-projectile mass ratio, impact velocity and impact angle. For asteroids, we developed a new approach which uses ML to train, validate, and test a neural network representation of the thermophysical behavior of the atmosphereless body. We applied the method to ground-based infrared observations of asteroid (25143) Itokawa and simulated infrared remote sensing of asteroid (101955) Bennu. We successfully retrieved surface roughness, thermal inertia of the regolith and rock component, and relative rock abundance. We foresee the application of the proposed methodology to support the forthcoming selection of the sampling site for future missions, such as NASA’s OSIRIS-REx mission and JAXA’s Hayabusa 2.
Wed Dec 5, 201811:00 am Sarah Hörst Johns Hopkins University Bystander Intervention Training
Tue Feb 19, 201911:00 am Hold for Possible Speaker