AMS Seminar: Laurent Younes (JHU- Applied Math & Stats) @ Whitehead 304

Title: Diffeomorphic Learning
Abstract: The talk introduces a learning paradigm in which the training data is transformed by a diffeomorphic transformation before prediction. The learning algorithm minimizes a cost function evaluating the prediction error on the transformed training set penalized by the distance between the diffeomorphism and the identity. The approach borrows ideas from shape analysis, in the way diffeomorphisms are estimated for shape and image alignment, to place them in a mostly unexplored setting, estimating, in particular diffeomorphisms in much larger dimensions. After introducing the concept and describing a learning algorithm,  diverse applications will be presented, mostly with synthetic examples, demonstrating the potential of the approach.