Home
About
Publications
Talks

Danny Tarlow

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

dtarlow cs toronto edu

News

I am again co-organizing a workshop on "Perturbations, Optimization, and Statistics" at NIPS 2014. See the webpage for the call for papers and more info.
I accepted a position as Researcher at Microsoft Research Cambridge. I'll keep this website up to date for now.
I am again co-organizing a workshop on "Perturbations, Optimization, and Statistics" at NIPS 2013. See the webpage for the call for papers and more info.
I started a Postdoc at Microsoft Research Cambridge. I'll keep this website up to date for now.
I am co-organizing a workshop on "Perturbations, Optimization, and Statistics" at NIPS 2012. The Call for Papers is now up.

Refereed Papers

A* Sampling (2014)
Chris Maddison, Daniel Tarlow, and Tom Minka.
Advances in Neural Information Processing Systems (NIPS 2014).
Oral Presentation
Outstanding Paper Award
[pdf] [appendix]

Just-In-Time Learning for Fast and Flexible Inference (2014)
S. M. Ali Eslami, Daniel Tarlow, Pushmeet Kohli, and John Winn.
Advances in Neural Information Processing Systems (NIPS 2014).
[pdf] [appendix]

Structured Generative Models of Natural Source Code (2014)
Chris Maddison and Daniel Tarlow.
The 31st International Conference on Machine Learning (ICML 2014).
[pdf]

Empirical Minimum Bayes Risk Prediction: How to extract an extra few % performance from vision models with just three more parameters (2014)
Vittal Premachandran, Daniel Tarlow, and Dhruv Batra.
The 27th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2014).
[pdf] [project]

Knowing what we don't know in NCAA Football ratings: Understanding and using structured uncertainty (2014)
Daniel Tarlow, Thore Graepel, and Tom Minka.
The 2014 MIT Sloan Sports Analytics Conference (SSAC 2014).
[pdf]

Learning Structured Models with the AUC Loss and Its Generalizations (2014)
Nir Rosenfeld, Ofer Meshi, Daniel Tarlow, and Amir Globerson.
The 17th International Conference on Artificial Intelligence and Statistics (AISTATS 2014).
[pdf] [appendix]

Learning to Pass Expectation Propagation Messages (2013)
Nicolas Heess, Daniel Tarlow, and John Winn.
Advances in Neural Information Processing Systems (NIPS 2013).
[pdf]

Tighter Linear Program Relaxations for High Order Graphical Models (2013)
Elad Mezuman*, Daniel Tarlow*, Amir Globerson, and Yair Weiss.
The 29th Conference on Uncertainty in Artificial Intelligence (UAI 2013). *Equal contribution.
[pdf] [bibtex]

Stochastic k-Neighborhood Selection for Supervised and Unsupervised Learning (2013)
Daniel Tarlow, Kevin Swersky, Laurent Charlin, Ilya Sutskever, and Richard Zemel.
The 30th International Conference on Machine Learning (ICML 2013).
[pdf + appendix] [code] [bibtex]

Exploring Compositional High Order Pattern Potentials for Structured Output Learning (2013)
Yujia Li, Daniel Tarlow, and Richard Zemel.
The 26th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2013).
Oral Presentation
[pdf] [appendix] [bibtex]

Probabilistic n-Choose-k Models for Classification and Ranking (2012)
Kevin Swersky, Daniel Tarlow, Ryan Adams, Richard Zemel, and Brendan Frey.
Advances in Neural Information Processing Systems (NIPS 2012).
[pdf] [appendix] [bibtex]

Cardinality Restricted Boltzmann Machines (2012)
Kevin Swersky, Daniel Tarlow, Ilya Sutskever, Ruslan Salakhutdinov, Richard Zemel, and Ryan Adams.
Advances in Neural Information Processing Systems (NIPS 2012).
[pdf] [code] [bibtex]

Fast Exact Inference for Recursive Cardinality Models (2012)
Daniel Tarlow, Kevin Swersky, Richard Zemel, Ryan Adams, and Brendan Frey.
The 28th Conference on Uncertainty in Artificial Intelligence (UAI 2012).
[pdf + appendix] [code] [bibtex]

Revisiting Uncertainty in Graph Cut Solutions (2012)
Daniel Tarlow and Ryan Adams.
The 25th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2012).
[pdf + appendix] [bibtex]

Randomized Optimum Models for Structured Prediction (2012)
Daniel Tarlow, Ryan Adams, and Richard Zemel.
The 15th International Conference on Artificial Intelligence and Statistics (AISTATS 2012).
[pdf + appendix] [bibtex]

Structured Output Learning with High Order Loss Functions (2012)
Daniel Tarlow and Richard Zemel.
The 15th International Conference on Artificial Intelligence and Statistics (AISTATS 2012).
[pdf + appendix] [bibtex]

Big and Tall: Large Margin Learning with High Order Losses (2011)
Daniel Tarlow and Richard Zemel.
CVPR 2011 Workshop on Inference in Graphical Models with Structured Potentials.
[pdf] [bibtex]

Graph Cuts is a Max-Product Algorithm (2011)
Daniel Tarlow, Inmar Givoni, Richard Zemel, and Brendan Frey.
The 27th Conference on Uncertainty in Artificial Intelligence (UAI 2011).
Oral Presentation
Best Student Paper Award - Runner Up
[pdf] [arXiv version] [bibtex]

Dynamic Tree Block Coordinate Ascent (2011)
Daniel Tarlow, Dhruv Batra, Pushmeet Kohli, and Vladimir Kolmogorov.
The 28th International Conference on Machine Learning (ICML 2011).
[pdf] [code at Dhruv's page] [bibtex]

HOP-MAP: Efficient Message Passing with High Order Potentials (2010)
Daniel Tarlow, Inmar Givoni, and Richard Zemel.
The 13th International Conference on Artificial Intelligence and Statistics (AISTATS 2010)
[pdf] [code] [bibtex]

Learning Articulated Structure and Motion (2010)
David Ross, Daniel Tarlow, and Richard Zemel
International Journal of Computer Vision (IJCV).
[journal] [webpage] [bibtex]

Automatically Calibrating a Probabilistic Graphical Model of Building Energy Consumption (2009)
Daniel Tarlow, Andrew Peterman, Benedict Schwegler, and Christopher Trigg.
The 11th International Building Performance Simulation Association Conference on Building Simulation.
[pdf] [bibtex]

Flexible Priors for Exemplar-based Clustering (2008)
Daniel Tarlow, Richard Zemel, and Brendan Frey.
The 24th Conference on Uncertainty in Artificial Intelligence (UAI 2008).
Oral Presentation
[pdf] [bibtex]

Unsupervised Learning of Skeletons from Motion (2008)
David Ross, Daniel Tarlow, and Richard Zemel.
The 10th European Conference on Computer Vision (ECCV 2008).
[pdf] [webpage] [bibtex]

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

Using Combinatorial Optimization within Max-Product Belief Propagation (2006)
John Duchi, Daniel Tarlow, Gal Elidan, and Daphne Koller.
Advances in Neural Information Processing Systems (NIPS 2006).
Spotlight Presentation
[pdf] [bibtex]

Working Papers

Detecting Parameter Symmetries in Probabilistic Models
Robert Nishihara, Tom Minka, and Daniel Tarlow.
[arXiv]

Course Projects, Tech Reports, Tutorials, etc.

Adaptive Tree CPDs in Max-Product Belief Propagation (2006)
Daniel Tarlow.
Course project for CSC2515, taught by Sam Roweis.
[pdf]


Press

Good Luck With That 'Perfect' March Madness Bracket. You'll Need It .
By Adam Cole, NPR Morning Edition, March, 2013.

Basketball and big data: Are robots the secret to winning your March Madness pool? .
By David Holmes, PandoDaily, March, 2013.

Who (Or What's) Best At Predicting March Madness Winners?.
By David Holmes, Fast Company, March, 2012.

Software to predict 'March Madness' basketball winner.
By MacGregor Campbell, New Scientist, March, 2011.

Smarter Buildings.
By Dan Falk, University of Toronto Magazine, March, 2010.


Around the Web


Selected Talks

Designing Loss Functions for Structured Prediction
Invited talk at Structured Prediction: Tractability, Learning, and Inference, Portland, Oregon. Summer 2013.

Fast Exact Inference with Cardinality-like Potentials
Hebrew University Learning Group Seminar, Jerusalem, Israel. Summer 2012.

Efficient Machine Learning with Combinatorial and High Order Structures
Microsoft Research Cambridge, Cambridge, UK. Spring 2012.

Graph Cuts is a Max-Product Algorithm
Uncertainty in Artificial Intelligence, Barcelona, Spain. Summer 2011.

Dynamic Tree Block Coordinate Ascent
International Conference on Machine Learning, Bellevue, Washington. Summer 2011.
[pptx] [pdf]

Learning with High Order Models and Loss Functions
Toronto Machine Learning Group Seminar, Toronto, Canada, Fall 2010.

Robust and Efficient Schedules for Chain and Grid Models in Infer.NET
Microsoft Research Cambridge, Cambridge, UK, Summer 2010.

On the Relationship Between Graph Cuts and Max-Product Belief Propagation
CBL Lab, University of Cambridge, Cambridge, UK, Summer 2010.

Max-Product Belief Propagation in High Order Factor Graphs
Toronto Machine Learning Group Seminar, Toronto, Canada, Fall 2009.
Microsoft Research Cambridge, Cambridge, UK, Summer 2009.

Automatically Calibrating a Probabilistic Graphical Model of Building Energy Consumption
IBPSA Conference on Building Simulation, Glasgow, Scotland, Summer 2009.

Flexible Priors for Exemplar-based Clustering
Uncertainty in Artificial Intelligence, Helsinki, Finland, Summer 2008.
Toronto Machine Learning Group Seminar, Toronto, Canada, Winter 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 Seminar, Toronto, Canada, Fall 2006.
[ppt] [pdf]

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

Partition-based Inference in Markov Networks
with John Duchi.
Daphne Koller group, Stanford, California, Spring 2006.

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

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


Other Stuff

Keyboard Shortcuts for Google Search in Your Browser Searchbox
Google labs has a product that lets you navigate search results using keyboard shortcuts. If you want to use keyboard shortcut Google search in your browser searchbox, you need to create an OpenSearch XML file describing the searchbox and ask the browser to load it. I did that here.

Update: Sadly, Google has killed this product.

- My take on March Madness predictions. (2011 Tournament Bracketology)

- My dad, who you should go see if you need knee surgery in Phoenix, AZ.