Jakob Foerster
Oxford associate professor and research scientist at FAIR (Meta); first author of COMA for multi-agent RL
- Associate Professor, Department of Engineering Science, University of Oxford1
- Research Scientist, FAIR, Meta AI, Meta1
- Founding Director, British Open-ended Learning and Discovery Lab (BOLD)1
Jakob N. Foerster is an Associate Professor in the Department of Engineering Science at the University of Oxford and a Research Scientist at FAIR, Meta AI; Foerster is also founding director of the BOLD (British Open-ended Learning and Discovery) lab. Foerster's work centers on multi-agent reinforcement learning: Foerster is first author of "Counterfactual Multi-Agent Policy Gradients" (COMA, 2017) and a co-author of QMIX (2018), both with Shimon Whiteson.123
Work
Counterfactual Multi-Agent Policy Gradients (COMA) (external site: arxiv.org)Published research
First author of the 2017 paper proposing COMA, a multi-agent actor-critic method that uses a centralised critic to learn decentralised policies.2
QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning (external site: arxiv.org)Published research
Co-author of the 2018 paper on factorising joint action-values for decentralised multi-agent execution.3
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.