4 Papers Accepted at ICLR 2021 and AAAI 2021

We are pleased to announce three accepted papers to ICLR and one paper at AAAI:ICLR 2021 : Transient Non-stationarity and Generalisation in Deep Reinforcement LearningMaximilian Igl, Gregory Farquhar, Jelena Luketina, Wendelin Boehmer, Shimon WhitesonICLR 2021 : My Body is a Cage: the Role of Morphology in Graph-Based Incompatible ControlVitaly Kurin, Maximilian Igl, Tim Rocktäschel, Wendelin Boehmer, Shimon WhitesonICLR 2021 : RODE: Learning Roles to Decompose Multi-Agent TasksTonghan Wang, Tarun Gupta, Anuj Mahajan, Bei Peng, Shimon Whiteson, Chongjie ZhangAAAI 2021 [...]

3 Papers Accepted at NeurIPS 2020

We are pleased to announce three accepted papers to NeurIPS:Can Q-Learning with Graph Networks Learn a Generalizable Branching Heuristic for a SAT Solver?Vitaly Kurin, Saad Godil, Shimon Whiteson, Bryan Catanzaro Learning Retrospective Knowledge with Reverse Reinforcement LearningShangtong Zhang, Vivek Veeriah, Shimon WhitesonWeighted QMIX: Improving Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement LearningTabish Rashid, Gregory Farquhar, Bei Peng, Shimon Whiteson

4 Papers Accepted at ICML 2020

We are pleased to announce four accepted papers to ICML:Growing Action SpacesGregory Farquhar, Laura Gustafson, Zeming Lin, Shimon Whiteson, Nicolas Usunier, Gabriel SynnaeveGradientDICE: Rethinking Generalized Offline Estimation of Stationary ValuesShangtong Zhang, Bo Liu, Shimon WhitesonProvably Convergent Two-Timescale Off-Policy Actor-Critic with Function ApproximationShangtong Zhang, Bo Liu, Hengshuai Yao, Shimon WhitesonDeep Coordination GraphsWendelin Boehmer, Vitaly Kurin, Shimon Whiteson

1 Paper Accepted to UAI

We are pleased to announce that “Multitask Soft Option Learning” is accepted to UAI 2020! Congratulations to the authors Maximilian Igl, Andrew Gambardella, Jinke He, Nantas Nardelli, N. Siddharth, Wendelin Böhmer, and Shimon Whiteson.The paper can be found here:

AAMAS Best Paper Award

“Deep Residual Reinforcement Learning” received the best paper award at AAMAS 2020! Congratulations to the authors Shangtong Zhang, Wendelin Boehmer, and Shimon Whiteson.The paper revisits Baird’s residual algorithm in deep RL. It stabilizes off-policy learning if equipped with bi-directional target net and alleviates distribution mismatch in Dyna planning.Paper:

2 Papers Accepted to AAMAS

We are happy about two full AAMAS papers with WhiRL members!“Deep Residual Reinforcement Learning”Shangtong Zhang, Wendelin Boehmer, Shimon Whiteson“Maximizing Information Gain via Prediction Rewards”Yash Satsangi, Sungsu Lim, Shimon Whiteson, Frans Oliehoek, Martha White

2 papers accepted to ICLR

We are very excited about two accepted ICLR 2020 papers and look forward to discussing our work in Ethiopia!“Optimistic Exploration even with a Pessimistic Initialisation” – Tabish Rashid, Bei Peng, Wendelin Boehmer, Shimon Whiteson“VariBAD: A Very Good Method for Bayes-Adaptive Deep RL via Meta-Learning” – Luisa Zintgraf, Kyriacos Shiarlis, Maximilian Igl, Sebastian Schulze, Yarin Gal, Katja Hofmann, Shimon WhitesonCongratulations also to new WhiRL member Kristian who is on two accepted papers with his old lab, including his first-author paper“Dynamical Distance Learning for Semi-Supervised and Unsupervised Skill Discovery” [...]

8 papers accepted to NeurIPS

We are excited about 8 accepted papers with WhiRL members, and look forward to discussing our work at NeurIPS 2019 in Vancouver!“Generalized Off-Policy Actor-Critic” – Shangtong Zhang, Wendelin Boehmer, Shimon Whiteson (“DAC: The Double Actor-Critic Architecture for Learning Options” – Shangtong Zhang, Shimon Whiteson (“Fast Efficient Hyperparameter Tuning for Policy Gradient Methods” – Supratik Paul, Vitaly Kurin, Shimon Whiteson (“VIREL: A Variational Inference Framework for Reinforcement Learning” – Matthew Fellows, Anuj Mahajan, Tim G. J. Rudner, Shimon Whiteson (Spotlight) (“MAVEN: [...]

IJCAI Survey: Reinforcement Learning Informed by Natural Language

To be successful in real-world tasks, Reinforcement Learning (RL) needs to exploit the compositional, relational, and hierarchical structure of the world, and learn to transfer it to the task at hand. Recent advances in representation learning for language make it possible to build models that acquire world knowledge from text corpora and integrate this knowledge into downstream decision making problems. We thus argue that the time is right to investigate a tight integration of natural language understanding into RL [...]

Vacancy: PostDoc

Together with Katja Hofmann (Microsoft Cambridge), we are hiring a postdoc for a special joint position in WhiRL at the University of Oxford and Microsoft Research Cambridge. More information: