His Ph.D. thesis is titled “Topics in Regularization and Boosting” and is concentrated around the importance of regularization — that is, controlling the complexity of prediction models — in modern data analysis, machine learning and data mining.