Who am I?
As of 2015, I am now a research scientist at Google DeepMind. My personal website has moved here.
I am a PhD Candidate in Neurobiology and Behavior at Columbia University under the supervision of Liam Paninski in the Department of Statistics. I am part of the Center for Theoretical Neuroscience here at Columbia, a wonderful group of people building new mathematical models of neurons and neural networks.
Broadly, my research interests are in machine learning techniques for rich, structured, sequential data, such as neural spike trains, natural language and audio. I am also interested in ideal observer models of how we might interpret the sort of complex stimuli that come naturally to us but computers struggle with.
My research in the Paninski group focuses on machine learning approaches for neural data analysis. We are currently looking at more robust methods for learning to decode motor actions from neural activity. We are also interested in models of how motor learning takes place, and how to use those models to augment data analysis.
I have also worked extensively with Frank Wood on nonparametric Bayesian modeling and inference. We focused on applications to prediction of natural language and other complex discrete sequences, specifically ways to make more compact models with equal predictive power.