A1149
Title: Can AI learn distributional regression
Authors: Brian Caffo - Johns Hopkins University (United States) [presenting]
Bonnie Smith - Johns Hopkins University (United States)
Abstract: The challenges in having artificial intelligence in the form of deep learning is considered in learning invariances associated with distributional regression. Distributional regression is a particular challenge to learn in an automated fashion since the assumption of exchangeability of the covariate is factorial in the number of invariances. The cost of attempting to learn invariances versus an alternative strategy of assuming potential invariances is considered. Applications to biomedical data is used to illustrate results.
