UAI 2014 - Probabilistic Inference Competition
We are pleased to announce an upcoming evaluation of probabilistic inference algorithms, as part of the UAI 2014 conference. All researchers working on inference in probabilistic graphical models are encouraged to participate. This will be the 5th UAI competition in a series of competitions that are organized every two years as part of UAI (the first evaluation was conducted in 2006).
Details about the competition and how to submit your inference solver(s) can be found on the competition webpage. At UAI 2014, winners will be invited to give 7-10 minute presentations about their solver and their approach.
Solver Submission: June 15th, 2014 - July 15th, 2014.
(You can submit your solvers multiple times between these two dates. We will only consider your latest version. Preliminary results and rankings of your solver will be available on the competition webpage given above)
- Results and Winners Announcement: July 19th, 2014
The evaluation will focus on the following computational tasks:
- Computing the partition function and probability of evidence (PR inference)
- Computing the marginal probability distribution over a variable given evidence (MAR inference)
- Computing the most likely assignment to all variables given evidence (MAP Inference)
New this year: The following two evaluations/competitions will be run on experimental basis:
- Computing the most likely assignment to a subset of variables given evidence (Marginal MAP inference)
- Weight (or Parameter) learning. We will use solvers that compute marginal probabilities (soft EM) and MAP assignments (hard EM) to do this. Therefore, if you participate in the MAR inference task or the MAP inference, your submission will be automatically entered in this task.
Submitted algorithms can compete in some or all of these tasks.
Feel free to email Vibhav Gogate (email@example.com) if you have any question.