About
Publications
Talks
Recent Bookmarks

Danny Tarlow

Ph.D. Student
University of Toronto
Dept. of Computer Science
Machine Learning Group

dtarlow cs toronto edu

About

I'm a Ph.D. student in the Machine Learning Group within the Department of Computer Science at the University of Toronto, where I also received my Master's Degree (M.Sc.) in January of 2008. I graduated from Stanford in June of 2006 with a Bachelor's Degree (BS) in Computer Science. My advisor is Richard Zemel, and I also work closely with Brendan Frey.

I'm broadly interested in the representation, reasoning and learning needed to apply probabilistic models to hard real-world problems. Practically, I rely on things like message passing algorithms, matrix factorizations, combinatorial optimization algorithms, and nonparametric Bayesian methods to provide motivation for how to approach a problem.

Beyond the practice, I am curious at a high theoretical level about why we choose the models that we do. What tradeoffs and implicit assumptions are we making when we choose a specific problem formulation? What analysis needs to be done so that we can justify or improve these decisions?

Most recently, I have been working on (and learning about) the following areas:

  • Discrete approximation algorithms.
  • Learning additional structure for collaborative filtering.
  • Message passing algorithms for nonparametric Bayesian models.


Peer Reviewed Publications

Unsupervised Learning of Skeletons from Feature Trajectories (2008)
with David Ross and Richard Zemel.
European Conference on Computer Vision (ECCV).
[pdf]

Flexible Priors for Exemplar-based Clustering (2008)
with Richard Zemel and Brendan Frey.
Uncertainty in Artificial Intelligence (UAI).
[pdf] [bibtex]

Learning Articulated Skeletons From Motion (2007)
with David Ross and Richard Zemel.
Workshop on Dynamical Vision at International Conference on Computer Vision (WDV-ICCV).
[pdf] [webpage] [bibtex]

Using Combinatorial Optimization within Max-Product Belief Propagation (2007)
with John Duchi, Gal Elidan, and Daphne Koller.
Advances in Neural Information Processing Systems (NIPS).
[pdf] [bibtex]


Selected Talks

Flexible Priors for Exemplar-based Clustering
Uncertainty in Artificial Intelligence, Helsinki, Finland, Summer 2008.
[pdf] [video]

Learning Articulated Skeletons from Motion
CIFAR Summer School on Neural Computation and Adaptive Perception, Summer 2007.
[pdf]

Using Combinatorial Optimization within Max-Product Belief Propagation
Toronto Machine Learning Group, Fall 2006.
[ppt] [pdf]

The Role of Features in a Feedback-based Ranking System
Google TechTalk, Summer 2006.

Partition-based Inference in Markov Networks
with John Duchi.
DAGS, Spring 2006.

Learning in General Games
with Lee Zen and Ankit Garg.
Stanford Logic Group, Winter 2005.
[link]

Automated Grading of Logic-based Homework Problems
Stanford Logic Group, Summer 2004.