cs 228 final project: exploring pac-bayesian theory

For my final project for CS 228: Computational Learning Theory (taught by Leslie Valiant), I wrote a report on PAC-Bayesian theory, which focuses on deriving generalization bounds (akin to those in the classical PAC learning framework) for Bayesian learning algorithms. This work is largely a survey of existing results and research directions, as well as initial directions exploring data-dependent priors formulated as output constraints.