While the amount of data that we store and consume is consistently growing, the similar trend is visible in the scale of the modern machine learning (ML) algorithms. Fueled by big data, these algorithms use many parameters to capture the intricate latent structure in the data. Hence, data continues to fuel the success of machine learning. In this talk, we discuss some practical ML models such as convolutional neural networks and linear dynamical systems. We provide insights and principled algorithms for rigorously learning these models from near-optimal amount of data. We also highlight how practical ideas and theoretical insights nicely coincide for some of these problems.

Prof. Samet Oymak