Title: Geometric inference in admixture models
Authors: Yuekai Sun - University of Michigan (United States) [presenting]
Abstract: A class of admixture models suitable for a variety of data types is considered. Fast and provably accurate inference algorithms are developed by accounting for the model's convex geometry and low dimensional simplicial structure. Thanks to the strong connection to the Voronoi tessellation and properties of the Dirichlet distribution, the proposed inference algorithm is shown to achieve consistency and strong error bound guarantees on a range of model settings and data distributions. The effectiveness of our model and the learning algorithm is demonstrated by simulations and by analyses of text and financial data.