Exploring the Frontiers of Deep Learning for Earth System Observation and Prediction

Abstract:

In the last few years, the Earth System Science community has rapidly come to adopt machine learning as a viable and useful approach for doing science, and it has been applied to a surprisingly diverse array of problems, with an ever-increasing degree of success. In its current form, ML is best viewed as an alternative and complimentary approach for software development. However, ML differs significantly in the capabilities it can build, the errors it produces, and the techniques needed for development and debugging. Furthermore, there are many ML challenges specific to science that need to be addressed including: massive data labelling, enforcing physical constraints, uncertainty quantification, explainability, reliability, AI safety, data movement problems, and the need for targeted benchmarks. Finally, it is my opinion that ML has the potential to grow far beyond its current limits, as a broad range of new possibilities become available when both the software and the software-engineer are composed of code. In this presentation, we will explore these issues and then survey cutting-edge research that is taking place on the frontiers of ML and the Earth System Sciences including: self-supervision, continual learning, online-learning, human in the loop, AutoML, neural architecture search, expanded use of GANs, dynamics loss-functions, spatio-temporal prediction, equation identification, ODE learning, differential programming, and more.