---- dataentry project ----
title :  Reinforcement Learning Competition 
contactname: Masoud Alipour
contactmail_mail: masoud.alipour@epfl.ch
contacttel: 021 6937529
contactroom: BC 150
type : bachelor semester 
state : unavailable
created_dt : 2009-06-03
taken_dt :  2010-12-01
completed_dt : 
by : 
output_media : 
table : projects
======
template:datatemplates:project
----

[[http://2009.rl-competition.org/|RL competition]] is a competition organized by [[http://www.cs.mcgill.ca/~icml2009/|International Conference on Machine Learning (ICML)]] committee to provide a platform for comparing the performance of Reinforcement Learning 
methods on a suite of challenging [[http://2009.rl-competition.org/domains.php|domains]] such as balancing a two-jointed virtual gymnast or controling a helicopter which is attempting to stably hover. 
Although the deadline for 2009 competition has already passed you may be able to participate to 2010 competition. Selected participant will be able to present their work at ICML. Note that you don't need to develop your own RL algorithm, you can start implementing an existing method and then try to improve it. The proportion of theoretical and empirical (programming, simulation) work in this project is 60-40 percent. 

Requirement:
  * Good programming skill 
  * Interest in AI and competition!
What will you learn
  * [[http://en.wikipedia.org/wiki/Reinforcement_learning|Reinforcement learning]]
  * [[http://en.wikipedia.org/wiki/Markov_decision_process|Markov decision processes]]
  * Optimizing your code to improve performance