Title: Shape analysis of points in the streaming model Speaker: Md. Ashraful Alam, UMass Amherst Abstract: Application of streaming algorithms to geometric problems is quite a new phenomenon. Most of the geometric problems involve point sets and all existing algorithms to solve such problems become tedious, if not impossible, when the point set is very large. For example, reconstructing objects from point clouds involves handling of massive point sets. Streaming algorithms may become very useful in such cases. In this talk, we explain the application of the streaming algorithms to analyze the shape of a large point set. We consider the alpha-hull and alpha-shape as geometric tools for analyzing the shape of a point set. We introduce the streaming algorithms to compute the alpha-hull and alpha-shape in dimension 2. Our algorithm computes the approximation of these shapes using only a fraction of the original point set. The error measure, which is the maximum distance from the exact shape to the approximate shape, depends on the number of points stored. If a large number of points are stored, then the error becomes very small and vice versa. To the best of our knowledge, there was no previous algorithm to compute the alpha-hull or alpha-shape in the streaming model.