Closed rogerkuou closed 3 years ago
Discussed this with Xu and Sonja. Xu has some new ideas of using RL to replace the current supervised learning + data assimilation approach. It seem to be feasible to some degree. During the discussion we get clear what he wants to do. But he needs to discuss this with Susan first because this makes a fundamental change to our road map.
Xu proposed too papers of RL to discuss.
de Witt, Hornigold - 2019 - Stratospheric Aerosol Injection as a Deep Reinforcement Learning Problem.pdf
Potentially he wants to see if the similar idea can be applied to the Vegetation project. As the PO of vegetation I will read the paper, discuss with Xu, and have a internal discussion with Sonja.