Time: Tuesday, Nov 1, 2016, 2:00pm @ MTH1311

Speaker: Dr. Alexander Cloninger (Yale University)

Title: Function Weighted Diffusion for Unreliable Functions with Applications to Determining Drug Treatment Effectiveness

Abstract: We consider building a function adapted diffusion operator high dimensional data X when the function F can only be evaluated on large subsets of the data, and possibly only depends on a small subset of the features. Our method breaks X up into hierarchical trees, and determines the importance of each feature in each subset of the population. The resulting metric is then used to define a localized filtration of F and estimation values of F at a finer scale than it is reliable naively. We apply this method in several cases in which F is unreliable a priori, and specifically use it to determine personalized risk and treatment effectiveness in drug trials. We validate the model on several synthetic datasets. Back to seminar