Resume

• 2010

  Google Research

  Applied machine vision methods to solve content-based video classification problems. Contributed\nto the DistBelief framework

  a close-to-the-metal distributed deep learning framework.

  Google Research

  Max Planck Institute for Intelligent Systems

  Tübingen

  Germany

  Visiting Researcher

  Doctor of Philosophy (PhD)

  Machine Learning

  University of Cambridge

• 2008

  French

  Master of Science (M.Sc.)

  Computer Science

  Green College

  The University of British Columbia / UBC

• 2006

  Invenia

  Co-founded a machine learning research consulting company. Recruited

  trained and supervised five\nresearch assistants

  plus consultants. Drafted

  presented and was awarded several research funding\ngrants. Led two research contracts applying modern machine learning methods to energy forecasts.\nThese projects led to the successful deployment of several automated forecasting systems for major\nutilities

  forecasting electric load

  wind generation and energy prices.

  Invenia

  Postodctoral Fellow

  Harvard University

  University of Toronto

  Assistant Professor

  Toronto

  Canada Area

• 2004

  Frantic Films

  University of Toronto

  Frantic Films

• Mathematical Modeling

  Science

  Pattern Recognition

  Probabilistic Models

  Python

  Machine Learning

  Data Analysis

  Software Engineering

  Statistics

  Matlab

  Research

  Consulting

  Artificial Intelligence

  Computer Science

  Algorithms

  Solid Presentation Skills

  LaTeX

  Optimally-Weighted Herding is Bayesian Quadrature

  We prove several connections between an efficient procedure for estimating moments (herding) which minimizes a worst-case error

  and a model-based way of estimating integrals (Bayesian Quadrature). It turns out that both are optimizing the same criterion

  and that Bayesian Quadrature is doing this in an optimal way. This means

  among other things

  that we can place worst-case error bounds on the optimal Bayesian estimator!

  Optimally-Weighted Herding is Bayesian Quadrature

  Zoubin Ghahramani

  Josh Tenenbaum

  Roger Grosse

  To search through an open-ended class of structured

  nonparametric regression models

  we introduce a simple grammar which specifies composite kernels. These structured models often allow an interpretable decomposition of the function being modeled

  as well as long-range extrapolation. Many common regression methods are special cases of this large family of models. We give several example decompositions time series.

  Structure Discovery in Nonparametric Regression through Compositional Kernel Search

  Zoubin Ghahramani

  Tomoharu Iwata

  If you fit a mixture of Gaussians to a single cluster that is curved or heavy-tailed

  your model will report that the data contains many clusters! To fix this problem

  we simply warp a latent mixture of Gaussians to produce nonparametric cluster shapes. The low-dimensional latent mixture model summarizes the properties of the high-dimensional clusters (or density manifolds) describing the data.

  Nonparametric Clustering with Warped Mixture Models

  Duvenaud

  Harvard University

  Max Planck Institute for Intelligent Systems