EWRL13 (2016)

The 13th European Workshop on Reinforcement Learning (EWRL 2016)

Dates: December 3-4 2016

Location:  Pompeu Fabra University, Barcelona, Spain (co-located with NIPS)
Ramon Turró building (building number 13). C/ Ramon Turró, 1 – 08005 Barcelona


[description] [submission] [dates] [committees] [keynotes] [papers] [registration] [venue] [schedule] [poster sessions] [photos] [sponsors]


The 13th European workshop on reinforcement learning (EWRL 2016) invites reinforcement-learning researchers to participate in the newest edition of this world class event. We plan to make this an exciting meeting for researchers worldwide, not only for the presentation of top quality papers, but also as a forum for ample discussion of open problems and future research directions. EWRL 2016 will consist of 11+ invited talks, contributed paper presentations, discussion sessions spread over a two day period, and a poster session.

Reinforcement learning is an active field of research which deals with the problem of sequential decision making in unknown (and often) stochastic and/or partially observable environments. Recently there has been a wealth of both impressive empirical results, as well as significant theoretical advances. Both types of advances are of significant importance and we would like to create a forum to discuss such interesting results.

The workshop will cover a range of sub-topics including (but not limited to):

  • Exploration/Exploitation and multi-armed bandits
  • Deep RL
  • Representation learning for RL
  • Large-scale RL
  • Theoretical aspects of RL
  • Policy search and actor-critic methods
  • Online learning algorithms
  • RL in non-stationary environments
  • Risk-sensitive RL
  • Transfer and Multi-task RL
  • Empirical evaluations in RL
  • Kernel methods for RL
  • RL in partially observable environments
  • Imitation learning and Inverse RL
  • Bayesian RL
  • Multi agent RL
  • Applications of RL
  • Open problems

Invited Speakers

Accepted Papers

Paper Submission

We invite submissions for the 13th European Workshop on Reinforcement Learning (EWRL 2016) from the entire reinforcement learning spectrum. Authors can submit a 2-6 pages paper in JMLR format (excluding
references) that will be reviewed by the program committee in a double-blind procedure. The papers can present new work or give a summary of recent work of the author(s). All papers will be considered for the poster sessions. Outstanding long papers (4-6 pages) will also be considered for a 20 minutes oral presentation. Accepted papers are going to be published in an arxiv.org collection.

Important Dates

  • Paper submissions due: 16/09/2016
  • Notification of acceptance: 06/10/2016
  • Camera ready due: 11/11/2016
  • Workshop begins: 03/12/2016
  • Workshop ends: 04/12/2016

Organizing Committee

Program Committee

  • Christos Dimitrakakis
  • Marc Bellemare
  • Christian Daniel
  • Marc Deisenroth
  • Amir-massoud Farahmand
  • Victor Gabillon
  • Matthieu Geist
  • Mohammad Ghavamzadeh
  • Mohammad Gheshlaghi Azar
  • Nan Jiang
  • Anders Jonsson
  • Akshay Krishnamurthy
  • Tor Lattimore
  • Alessandro Lazaric
  • Ashique Rupam Mahmood
  • Timothy Mann
  • Jérémie Mary
  • Rémi Munos
  • Laurent Orseau
  • Ronald Ortner
  • Ian Osband
  • Bilal Piot
  • Doina Precup
  • Marcello Restelli
  • Scott Sanner
  • Georgios Theocharous
  • Michal Valko

Keynote/Tutorial/Invited Speakers’ Abstracts



Please follow this link to register:
Registration is free of charge, but please note that we have limited space. Thus, we kindly ask you to register only if you are really planning to participate.

Workshop Venue

EWRL13 takes place at the Ciutadella campus of the Pompeu Fabra University, in Barcelona, Spain. The precise address is:

Universitat Pompeu Fabra, Ramon Turró building (building number 13)
C/ Ramon Turró, 1 – 08005 Barcelona

Workshop Schedule

Sat 3

8:30 – 8:40 Opening remarks
8:40 – 9:20 Invited talk: Bruno Scherrer (INRIA) — On Periodic Markov Decision Processes
9:20 – 9:40 Why is Posterior Sampling Better for RL? (Ian Osband)
9:40 – 10:00 Linear Thompson Sampling Revisited (Marc Abeille)

10:00 – 10:30 Coffee break

10:30 – 11:10 Invited talk: Héctor Geffner (Universitat Pompeu Fabra)
11:10 – 11:50 Invited talk: John Langford (Microsoft Research) — The Contextual Reinforcement Learning Research Program
11:50 – 12:10 A Deep Hierarchical Approach to Lifelong Learning in Minecraft (Tom Zahavy)
12:10 – 13:30 Poster session 1

13:30 – 15:00 Lunch break (on your own)

15:00 – 15:40 Invited talk: Remi Munos (Google DeepMind and INRIA) — Safe and Efficient Off-Policy Reinforcement Learning
15:40 – 16:00 Principled Option Learning in Markov Decision Processes (Michal Moshkovitz)
16:00 – 16:20 Exploration–Exploitation in MDPs with Options (Ronan Fruit)

16:20 – 16:50 Coffee break

16:50 – 17:30 Invited talk: Doina Precup (McGill University) — How to construct good temporal abstractions
17:30 – 18:10 Invited talk: Ronald Ortner (Montanuniversität Leoben) — Some open problems for average reward MDPs
18:10 – 19:00 Panel discussion 1

19:00 – 21:00 Dinner break (on your own)
21:00 – Event

Sun 4

9:00 – 9:20 Situational Awareness by Risk-Conscious Skills (Shie Mannor)
9:20 – 9:40 Memory Lens: How Much Memory Does an Agent Use? (Christoph Dann)
9:40 – 10:00 Continuous LSPI (Bilal Piot)

10:00 – 10:30 Coffee break

10:30 – 11:10 Invited talk: Mohammad Ghavamzadeh (Adobe Research and INRIA) — Learning Safe Policies in Sequential Decision-Making Problems
11:10 – 11:50 Invited talk: Alessandro Lazaric (INRIA) — Spectral Methods for Reinforcement Learning
11:50 – 12:10 Accelerated Gradient Temporal Difference Learning (Martha White)
12:10 – 13:30 Poster session 2

13:30 – 15:00 Lunch break (on your own)

15:00 – 15:40 Invited talk: Emma Brunskill (Carnegie Mellon University) — Helping unlock the potential of RL
15:40 – 16:00 Value-Aware Loss Function for Model Learning in Reinforcement Learning (A-M. Farahmand)
16:00 – 16:20 Consistent On-Line Off-Policy Evaluation (Assaf Hallak)

16:20 – 16:50 Coffee break

16:50 – 17:30 Invited talk: Sergey Levine (University of Washington and Google) — Challenges in Deep Reinforcement Learning
17:30 – 18:10 Invited talk: Gerhard Neumann (University of Lincoln) — Information-theoretic Policy Search Methods for Learning Versatile, Reusable Skills
18:10 – 19:00 Panel discussion 2

Poster sessions

Poster session 1 (Saturday):
Robust Kalman Temporal Difference (Shirli Di-Castro Shashua and Shie Mannor)
A Lower Bound for Multi-Armed Bandits with Expert Advice (Yevgeny Seldin and Gábor Lugosi)
Magical Policy Search: Data Efficient Reinforcement Learning with Guarantees of Global Optimality (Philip Thomas and Emma Brunskill)
Approximations of the Restless Bandit Problem (Steffen Grunewalder and Azadeh Khaleghi)
Bayesian Optimal Policies for Asynchronous Bandits with Known Trends (Mohammed Amine Alaoui, Tanguy Urvoy and Fabrice Clérot)
Exploration Potential (Jan Leike)
Corrupt Bandits (Pratik Gajane, Tanguy Urvoy and Emilie Kaufmann)
Iterative Hierarchical Optimization for Misspecified Problems (Daniel J. Mankowitz, Timothy Mann and Shie Mannor)
A Deep Hierarchical Approach to Lifelong Learning in Minecraft (Chen Tessler, Shahar Givony, Daniel J. Mankowitz, Tom Zahavy and Shie Mannor)
Situational Awareness by Risk-Conscious Skills (Daniel J. Mankowitz, Aviv Tamar and Shie Mannor)

Poster session 2 (Sunday):
Decoding multitask DQN in the world of Minecraft (Lydia Liu, Urun Dogan and Katja Hofmann)
Spatio-Temporal Abstractions in Reinforcement Learning Through Neural Encoding (Nir Baram, Tom Zahavy and Shie Mannor)
Non-Deterministic Policy Improvement Stabilizes Approximated Reinforcement Learning (Wendelin Böhmer, Rong Guo and Klaus Obermayer)
Automatic Representation for Life-Time Value Recommender Systems (Assaf Hallak, Elad Yom-Tov and Yishay Mansour)
Using Policy Gradients to Account for Changes in Behavior Policies under Off-policy Control (Lucas Lehnert and Doina Precup)
Deep Reinforcement Learning Solutions for Energy Microgrids Management (Vincent Francois-Lavet, David Taralla, Damien Ernst and Raphael Fonteneau)
Toward a data efficient neural actor-critic (Matthieu Zimmer, Yann Boniface and Alain Dutech)





%d bloggers like this: