A Perspective on Machine Learning

In this talk I will present an overview of my research with the hope of creating new research links and collaborations at Oxford. More specifically, I will cover the following topics:

(i) Machine learning techniques for massive scale industrial data applications, including Zite (which powers CNN trends) and KnoBout (a company I recently co-founded with the vision of mining all accessible reviews on the web). 

(ii) Structured probabilistic models for decision making and planning. 

(iii) Probabilistic methods for optimization and numerical computing, with application to automatic configuration of systems, hardware, interfaces, models and algorithms. 

(iv) Recent advances on adaptive Markov chain Monte Carlo inference. 

(v) Deep learning and the quest for automatic machine perception.