Pattern Theory

This course is an introduction to some probabilistic models and numerical algorithms that model some aspects of human cognition. The class begins with stochastic models of language introduced by Shannon and develops related models for speech and vision. The classes stresses mathematical foundations, in particular the role of information theory in developing Bayesian models and the increasing importance of dynamics in several algorithms, especially in optimization and deep learning. Student assessment will be based on computational projects that implement the principles discussed in lecture.