EWRL16 (2023)
The 16th European Workshop on Reinforcement Learning (EWRL 2023)
Dates: 14-16 September 2023
Location: Brussels, Belgium
Schedule: Add to calendar
Contact: ewrl2023@gmail.com
Social Event
Friday evening there will be a foodtruck event on VUB campus for all particpants. This is the perfect opportunity to socialise with other participants while enjoying some food and drinks. We will start at 19:00 and finish around 22:00 and you can use the tickets received in your welcome package to choose from a variety of meals and drinks. Please find the map marker to the location here.
Workshop Schedule
We have prepared a schedule that can be added to your calendar by clicking here.
Day 1 – Thursday 14 Sept
08:30-09:30 Registration and coffee
09:30-10:00 Welcome address
10:00-11:00 Invited talk – Emilie Kaufmann: An instance-dependent view on PAC Reinforcement Learning
11:00-11:30 Coffee
11:30-12:30 Invited talk – Michal Valko: Curious world models
12:30-14:00 Lunch
14:00-15:30 Posters and coffee
15:30-16:30 Invited talk – Julien Perolat: Mastering the game of Stratego with model-free multiagent reinforcement learning
16:30-18:00 Posters
Day 2 – Friday 15 Sept
09:00-11:00 Tutorial – Shie Mannor: Robust sequential decision making
11:00-12:30 Coffee break and posters
12:30-14:30 Lunch
14:30-15:30 Invited talk – Martha White: A New Policy Update in Actor-Critic Algorithms
15:30-16:30 Invited talk – David Silver: Beyond AlphaGo
16:30-18:00 Posters and coffee
19:00 – 22:00 Social event – Food trucks
Day 3 – Saturday 16 Sept
09:00-11:00 Tutorial – Olivier Pietquin: RL and Language: long story short
11:00-12:30 Posters and coffee
12:30-14:30 Lunch
14:30-15:30 Invited talk – Georg Martius: Intrinsic Motivation in Reinforcement Learning (Extra videos to accompany the presentation are available here)
15:30-16:00 Closing remarks
16:00-17:30 Posters and coffee
Invited Speakers
- Emilie Kaufmann Université de Lille
- David Silver Google DeepMind
- Martha White University of Alberta
- Julien Perolat Google DeepMind
- Shie Mannor Technion Israel Institute of Technology
- Georg Martius Max Planck Institute for Intelligent Systems
- Michal Valko Google DeepMind
- Olivier Pietquin Google DeepMind
Poster Sessions
All accepted papers are assigned a day when they may present their poster. The papers can be accessed through openreview by clicking on the respective title.
Note for presenters: The maximum size for posters is A0 in portrait mode.
Thursday – 14/09/2023
- Optimal Hierarchical Average-Reward Linearly-solvable Markov Decision Processes
- Cooperative Foraging Behaviour Through Multi-Agent Reinforcement Learning with Graph-Based Communication
- Value-Distributional Model-Based Reinforcement Learning
- Stateless Mean-Field Games: A Framework for Independent Learning with Large Populations
- Online Learning under Adversarial Nonlinear Constraints
- Train Hard, Fight Easy: Robust Meta Reinforcement Learning
- Reinforcement Learning in near-continuous time for continuous state-action spaces
- E-MCTS: Deep Exploration in Model-Based Reinforcement Learning by Planning with Epistemic Uncertainty
- Combinatorial Stochastic-Greedy Bandit
- Sample Complexity of Goal-Conditioned Hierarchical Reinforcement Learning
- Cancellation-Free Regret Bounds for Lagrangian Approaches in Constrained Markov Decision Processes
- Delayed Feedback in Generalised Linear Bandits
- Online Learning in Autoregressive Dynamics
- Interpretable (Un)Controllable Features in MDP’s
- Prioritizing States with Action Sensitive Return in Experience Replay
- Adversarial Distributional Reinforcement Learning against Extrapolated Generalization (Extended Abstract)
- Interaction of doctors with explainable RL decision support via behavioural readouts of eye-tracking
- Laser Learning Environment: Insights on the challenges of coordination-critical multi-agent tasks
- Beyond Markovian RL: Efficient Offline RL in Regular Decision Processes
- Least Squares Inverse Q-Learning
- Overcoming Policy Collapse in Deep Reinforcement Learning
- Constant or logarithmic regret in asynchronous multiplayer bandits
- On the Statistical Efficiency of Mean Field RL with General Function Approximation
- Online Decision Tree Construction with Deep Reinforcement Learning
- Beyond Average Reward in Markov Decision Processes
- A Gradient Critic for Policy Gradient Estimation
- Learning State Reachability as a Graph in Translation Invariant Goal-based Reinforcement Learning Tasks
- Optimistic Planning by Regularized Dynamic Programming
- Towards Faster Global Convergence of Robust Policy Gradient Methods
- The Role of Diverse Replay for Generalisation in Reinforcement Learning
- Kernelized Reinforcement Learning with Order Optimal Regret Bounds
- Integrating Distributed Architectures in Highly Modular RL Libraries
- Optimism & Adaptation in Policy Optimization
- On Imitation in Mean-field Games
- The Wasserstein Believer: Learning Belief Updates for Partially Observable Environments through Reliable Latent Space Models
- Bayesian Deep Q-Learning via Sequential Monte Carlo
- Wasserstein Auto-encoded MDPs: Formal Verification of Efficiently Distilled RL Policies with Many-sided Guarantees
- Proximal Point Imitation Learning
- Informed POMDP: Leveraging Additional Information in Model-Based RL
Friday – 15/09/2023
- APART: Diverse Skill Discovery using All Pairs with Ascending Reward and DropouT
- Fast Treatment Personalization with Latent Bandits in Fixed-Confidence Pure Exploration
- Understanding Deep Neural Function Approximation in Reinforcement Learning via $\epsilon$-Greedy Exploration
- Delayed Bandits: When Do Intermediate Observations Help?
- Interactive and Concentrated Differential Privacy for Bandits
- Delphic Offline Reinforcement Learning under Nonidentifiable Hidden Confounding
- Coherent Soft Imitation Learning
- AdaStop: sequential testing for efficient and reliable comparisons of Deep RL Agents
- Hyperparameters in Reinforcement Learning and How To Tune Them
- Efficient Dynamics Modeling in Interactive Environments with Koopman Theory
- Optimal Convergence Rate for Exact Policy Mirror Descent in Discounted Markov Decision Processes
- A Provably Efficient Option-Based Algorithm for both High-Level and Low-Level Learning
- Truncating Trajectories in Monte Carlo Policy Evaluation: an Adaptive Approach
- Online Adversarial MDPs with Off-Policy Feedback and Known Transitions
- Diverse Offline Imitation via Fenchel Duality
- Revisiting Continuous-Time Reinforcement Learning. A Study of HJB Solvers Based on PINNs and FEMs
- Mind the Uncertainty: Risk-Aware and Actively Exploring Model-Based Reinforcement Learning
- Multi-Armed Bandits with Generalized Temporally-Partitioned Rewards
- Continuous Episodic Control
- Stochastic Rising Bandits: A Best Arm Identification Approach
- Pure Exploration in Bandits with Linear Constraints
- Zero-shot stitching in Reinforcement Learning using Relative Representations
- Goal-conditioned Offline Planning from Curious Exploration
- On Preemption and Learning in Stochastic Scheduling
- Online Configuration in Continuous Decision Space
- Dual RL: Unification and New Methods for Reinforcement and Imitation Learning
- Two-Memory Reinforcement Learning
- Switching Latent Bandits
- Long-Distance Electric Vehicle Navigation using a Combinatorial Semi-Bandit Approach
- Representation-Driven Reinforcement Learning
- Colored Noise in PPO: Improved Exploration and Performance Through Correlated Action Sampling
- Provably Learning Nash Policies in Constrained Markov Potential Games
- Exploiting Causal Graph Priors with Posterior Sampling for Reinforcement Learning
- Online Regret Bounds for Satisficing in MDPs
- AutoRL Hyperparameter Landscapes
- Bad Habits: Policy Confounding and Out-of-Trajectory Generalization in RL
- Evolutionary Preference-Based Reinforcement Learning for Partially Observable Environments
- Automated Design of Affine Maximizer Mechanisms In Dynamic Settings
Saturday – 16/09/2023
- On the importance of data collection for training general goal-reaching policies
- Offline Primal-Dual Reinforcement Learning for Linear MDPs
- A Unified Perspective on Value Backup and Exploration in Monte-Carlo Tree Search
- Safe and Efficient Operation with Constrained Hierarchical Reinforcement Learning
- A Novel Framework for Policy Mirror Descent with General Parametrization and Linear Convergence
- Provable and Practical: Efficient Exploration in Reinforcement Learning via Langevin Monte Carlo
- First- and Second-Order Bounds for Adversarial Linear Contextual Bandits
- On the Complexity of Differentially Private Best-Arm Identification with Fixed Confidence
- Conditional Mutual Information for Disentangled Representations in Reinforcement Learning
- Robust Reinforcement Learning via Adversarial Kernel Approximation
- Iterated Deep Q-Network: Efficient Learning of Bellman Iterations for Deep Reinforcement Learning
- Skill or Luck? Return Decomposition via Advantage Functions
- Approximate information state based convergence analysis of recurrent Q-learning
- On-Demand Communication for Asynchronous Multi-Agent Bandits
- Adaptive Algorithms for Relaxed Pareto Set Identification
- Regularity as Intrinsic Reward for Free Play
- Percentile Criterion Optimization in Offline Reinforcement Learning
- A Patterns Framework for Incorporating Structure in Reinforcement Learning
- Lifelong Best-Arm Identification with Misspecified Priors
- Finite-state Offline Reinforcement Learning with Moment-based Bayesian Epistemic and Aleatoric Uncertainties
- An $\varepsilon$-Best-Arm Identification Algorithm for Fixed-Confidence and Beyond
- Robust Knowledge Transfer in Tiered RL
- On the Sample Complexity of Inverse Reinforcement Learning
- Submodular Reinforcement Learning
- Revisiting the Static Model in Robust Reinforcement Learning
- Diverse Projection Ensembles for Distributional Reinforcement Learning
- Regret Minimization via Saddle Point Optimization
- Distributional Policy Evaluation: a Maximum Entropy approach to Representation Learning
- Curiosity-driven Exploration in Sparse-reward Multi-agent Reinforcement Learning
- Probabilistic Inference in Reinforcement Learning Done Right
- Learning Hierarchical World Models with Adaptive Temporal Abstractions from Discrete Latent Dynamics
- Best Policy Identification in Discounted Linear MDPs
- Contextualize Me – The Case for Context in Reinforcement Learning
Workshop Venue
EWRL16 is hosted at the campus of the Vrije Universiteit Brussel (VUB) in the heart of Brussels, the beautiful capital of Europe. The exact address is: Pleinlaan 2, 1050 Elsene in Building I and you can find a maps marker at this link. The building itself may not show up on the map as it is fairly new.
Reaching the Venue
Brussels is well-connected and can be reached by car, train or aeroplane. The main airport is Brussels Airport located in the city of Zaventem, just outside of Brussels. Attendees who fly into this airport have the convenient option of taking a train directly inside the airport to the 3 main train stations Brussels Nord, Brussels Central and Brussels South (Midi), but also to Etterbeek, the station next to the VUB campus. There are also taxis and busses nearby. The low-fare airport Brussels South Charleroi Airport is located about an hour’s drive from Brussels city but you can book a shuttle bus in advance.
When travelling to Brussels by train, there are multiple train stops inside the city which are also connected to other public transport options such as bus, tram and metro. Most international trains stop at Brussels-Midi (Bruxelles-Midi in French). For any train travel within Belgium, we recommend checking the official website.
For attendees who plan to travel by car, we highlight that Brussels is a Low Emission Zone (LEZ). We advise you to check in advance whether your vehicle is allowed.
The workshop will be held on the VUB campus in Etterbeek inside building I. The campus is easily accessible using public transport. There is a train stop next to the university called “Etterbeek Station” and there are multiple bus and tram stops on all sides of the campus. There is also a metro stop “Petillon” approximately 10 minutes walking from the campus. All public transport operated in Brussels by the MIVB offers contactless payment with credit/debit card inside the vehicle. You can also buy a Brupass ticket for multiple rides. Google Maps is well integrated with the public transport schedule, making it easy to plan your trips in advance.
Note that the VUB has two locations in Brussels, with the other one being in Jette. Please make sure you come to Etterbeek as Jette is on the opposite side of Brussels. For more details and a complete campus plan, please see the official webpage.
Accommodations
We are aware of other events happening at the same time as the workshop, so we recommend booking accommodations as soon as possible. We suggest looking on platforms such as booking.com or Airbnb.
Organizing Committee
General Chair
- Ann Nowé Vrije Universiteit Brussel – Brussels, Belgium
Program Committee Chairs
- Emmanuel Rachelson ISAE-SUPAERO – Toulouse, France
- Ciara Pike-Burke Imperial College London – London, United Kingdom
- Andrea Tirinzoni Meta AI – Paris, France
- Dylan R. Ashley The Swiss AI Lab IDSIA (USI-SUPSI) – Lugano, Switzerland
Local Chairs
- Willem Röpke Vrije Universiteit Brussel – Brussels, Belgium
- Raphael Avalos Vrije Universiteit Brussel – Brussels, Belgium
Sponsors
Code of Conduct
The official EWRL 2023 Code of Conduct can be found here.
European Workshops on Reinforcement Learning 
You must be logged in to post a comment.