Statistical models and sheaves

Monday, May 21, 2018 - 11:30am - 12:30pm
Lind 305
Sayan Mukherjee (Duke University)
I will give a basic overview of statistical principles, both in terms of classical statistics and more modern machine learning perspectives. I will then discuss how sheaves can enter into statistical thinking both from a likelihood model based perspective as well direct models of data such as manifold learning. I will close with some concrete examples from Markov chain theory, modeling shapes and surfaces, as well as inference in dynamical systems.