Michael started his DPhil in Mathematics at Oxford in October 2017, supervised by Professor Jared Tanner. Prior to joining Oxford and the Turing he worked as a strategy consultant
Michael’s research is concerned with the mathematical analysis of algorithms used in machine learning. His general approach is to make certain structural assumptions on the data, e.g., sparse under a certain transform, arising as a result of a particular generative model or lying on some smooth manifold and analyse conditions under which various algorithms, at least with high probability, are successful. To do this he leverages a range of tools, techniques and ideas from a number of areas, including applied probability, numerical linear algebra, optimisation and discrete mathematics. He is particularly interested in how and why deep learning performs so well.