Joint workshop at ICML, UAI, and COLT 2009
 June 18, 2009, Montreal, Canada
Overview

Call for Papers

Accepted Papers

Program Committee

Submission Instructions

Important Dates

Program

Sponsors

Photo Album

ICML Workshop Video Lectures

 PROGRAM

 8:20 AM
 
Welcome
Özgür Şimşek & George Konidaris

Invited talk: High-level actions (watch)
Stuart Russell

 9:30 AM Paper presentations
Manifold embeddings for model-based reinforcement learning of neurostimulation policies (watch)
Keith Bush, Joelle Pineau and Massimo Avoli
Dyna(k): A multi-step Dyna planning (watch)
Hengshuai Yao, Rich Sutton, Shalabh Bhatnagar, Dongcui Diao and Csaba Szepesvári
An empirical comparison of abstraction in models of Markov decision processes (watch)
Todd Hester and Peter Stone
Towards feature selection in actor-critic algorithms
Khashayar Rohanimanesh, Nicholas Roy and Russ Tedrake

 10:30 AM Coffee break

 11:00 AM Paper presentations
Basis function construction for hierarchical reinforcement learning (watch)
Sarah Osentoski and Sridhar Mahadevan
Finding equivalences among abstract actions
Alicia P. Wolfe

 11:30 AM
 
Panel discussion: The future of abstraction in reinforcement learning
Panelists: Andrew Barto, Tom Dietterich (slides), Michael Littman and Ron Parr

 12:30 PM Poster spotlights (and lunch)
Improving Bayesian reinforcement learning using transition abstraction
Daniel Acuna and Paul Schrater
Situation dependent spatial abstraction in reinforcement learning based on structural knowledge (watch)
Lutz Frommberger
Automated discovery of options in factored reinforcement learning (watch)
Olga Kozlova, Olivier Sigaud and Christophe Meyer
Hierarchical skill learning for high-level planning (watch)
James MacGlashan and Marie desJardins
 
 1:30 PM
 
Invited talks by Rich Sutton, Wolfram Schultz, Gerald Tesauro and Andrew Ng
at the Multidisciplinary Symposium on Reinforcement Learning (MSRL)
 
  6:00 PM Joint poster session with MSRL