613 SEMINAR SERIES: Tianle Yuan

Affiliation: UMBC-JCET/Climate & Radiation Laboratory
Event Date: Wednesday, December 4, 2019

Location: Building 33, Room H114
Time: 1:30 PM

Combining Experts and Deep Learning to Obtain Insights from NASA Data

In this talk, we introduce the machine-expert hybrid approach to explore vast amounts of NASA data. Specifically, we propose to combine the power of experts with domain knowledge and deep learning models to explore NASA data. The approach can enable obtaining scientific insights more efficiently and in new avenues. We will present a few examples of using this approach to study marine low cloud morphology, find ship-tracks and study aerosol-cloud interactions, and parameterize sub-grid low clouds in a data-driven fashion. We show how the hybrid approach can scale up data exploration in terms of both depth and breadth. The resulting products and methods will enable research communities to tackle pressing research issues such as low cloud feedbacks, aerosol-cloud interactions and cloud modeling with new tools.

 

Seminar Series Coordinators

Yuekui.Yang-1@nasa.gov
Jae.N.Lee@nasa.gov