Using Artificial Intelligence for hypothesis testing and physical insight extraction

Artificial Intelligence techniques excel at predicting values or classifications based on complex training data. However, they do so as a "black box" and often offer no new insight on the physical processes behind the data. In this talk I will discuss uses of AI in which our focus is in obtaining a better understanding of the physical processes involved in generating the data in contrast to standard applications where the goal is minimizing the cost function. I will show research my group is doing on galaxy distance estimations, Large Scale Structure and galaxy formation.

Speaker: Miguel Aragon-Calvo, UNAM