Colloquium
Department of Physics, NCU
Small Atom and Big Data: Applications and Challenges of Machine Learning in Atomistic Scale Simulations
Prof. Chun-Wei Pao (包淳偉)
Research Center for Applied Sciences, Academia Sinica
Date 2020.03.17 (Tue)
Place S4-625
Time 13:30-15:00
Abstract:
Machine learning has drawn significant amount of attentions around the world because of their capability of building predictive models that are difficult to be extracted solely by human brains. In this talk, I will talk about application of machine learning models on atomistic simulation of complex material systems. I will briefly introduce the machine-learning-enabled energy predictors, and show their applications in complex perovskite materials and novel complex earth-abundant solar cell materials. I will demonstrate that the machine-learning-enabled models offer tens of thousand times computational speedup relative to first-principle calculations, and will also discuss the drawback of these models.