The NASA Frontier Development Lab is an AI research accelerator that was formed by means of a collaborative partnership between public and private entities, in order to tackle challenging problems relating to Earth Observations, Space Weather, Space Resources, Astrobiology and Exoplanets.
At the 2018 FDL, I was part of the Space Weather Challenges Team, charged with applying Artificial Intelligence/Machine Learning (AI/ML) techniques towards forecasting disruptions in GPS.
Clockwise from top left: With the Space Weather Team at the NVIDIA Endeavour campus; at the SETI institute in Mountain View, CA; presenting our final talk at the FDL Event Horizon; at NASA Ames.
Presentation slides available at this link, and a recording of the talk is available here.
Improving on existing baseline models that forecast ionospheric scintillation, we showed that significant improvements (~ 70% in performance metrics) in predictive capabilities could be achieved by expanding the input space and experimenting with various model architectures. We also showed the ionospheric scintillation, despite being driven by global-scale physical drivers, have distinctively localized behaviours that could be inferred from colocated magnetometer and scintillation measurements. As a result of this, our team won the "Unexpected Discovery Award".
The FDL team!