The Probabilistic Inference Challenge (PIC2011)

Overview

Probabilistic inference in graphical models is a key task in many applications, from machine vision to computational biology. Since the problem is generally computationally intractable many approximations have been suggested over the years.
The goal of this challenge is to evaluate inference algorithms on difficult large scale problems.
Algorithms for MAP, marginals, and partition function approximation will be evaluated. An online leaderboard will show the relative rankings of the algorithms.

Challenge Results

The challenge is now over. The winners are:
On the marginals and partition function tasks (20 seconds, 20 minutes and 1 hour): Vibhav Gogate, University of Texas, Dallas.
On the MAP task (20 seconds, 20 minutes and 1 hour): Lars Otten, Alexander Ihler, Kalev Kask, Rina Dechter, UC Irvine. Also using some open-source code by Frank Hutter, UBC.
Congratulations to the two winning teams! This was a very close race, as you can see from the leaderboard .

News

The protein-protein networks have been released (see Examples).

Provisional Timetable

July 15th, 2011: Site open to registration and submission of solvers.
December 20th, 2011: Challenge Begins! No new models added after this date.
March 15th, 2012: Site closed to submission
March 29th, 2012: Winners in different categories announced.
After the Challenge: The site will remain open for evaluation on the challenge datasets, and others that will be added.

PRIZES

An award of 100 Euro will be given for each of the nine challenge categories.

ORGANIZERS

Gal Elidan - Hebrew University
Amir Globerson - Hebrew University

STUDENT ORGANIZER AND WEB-MASTER

Uri Heinemann - Hebrew University

Acknowledgments

The organizers gratefully acknowledge support from the PASCAL Network of Excellence.

| Contact admin |