Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox
Picture for Jakob Foerster

Jakob Foerster

Replay-Guided Adversarial Environment Design


Oct 06, 2021
Minqi Jiang, Michael Dennis, Jack Parker-Holder, Jakob Foerster, Edward Grefenstette, Tim RocktÀschel

Figure 1 for Replay-Guided Adversarial Environment Design
Figure 2 for Replay-Guided Adversarial Environment Design
Figure 3 for Replay-Guided Adversarial Environment Design
Figure 4 for Replay-Guided Adversarial Environment Design
* NeurIPS 2021 
  Access Paper or Ask Questions

Don't Sweep your Learning Rate under the Rug: A Closer Look at Cross-modal Transfer of Pretrained Transformers


Jul 26, 2021
Danielle Rothermel, Margaret Li, Tim RocktÀschel, Jakob Foerster

Figure 1 for Don't Sweep your Learning Rate under the Rug: A Closer Look at Cross-modal Transfer of Pretrained Transformers
Figure 2 for Don't Sweep your Learning Rate under the Rug: A Closer Look at Cross-modal Transfer of Pretrained Transformers
Figure 3 for Don't Sweep your Learning Rate under the Rug: A Closer Look at Cross-modal Transfer of Pretrained Transformers
Figure 4 for Don't Sweep your Learning Rate under the Rug: A Closer Look at Cross-modal Transfer of Pretrained Transformers
* Accepted to ICML 2021 Workshop: Self-Supervised Learning for Reasoning and Perception 
  Access Paper or Ask Questions

Implicit Communication as Minimum Entropy Coupling


Jul 17, 2021
Samuel Sokota, Christian Schroeder de Witt, Maximilian Igl, Luisa Zintgraf, Philip Torr, Shimon Whiteson, Jakob Foerster

Figure 1 for Implicit Communication as Minimum Entropy Coupling
Figure 2 for Implicit Communication as Minimum Entropy Coupling
Figure 3 for Implicit Communication as Minimum Entropy Coupling
Figure 4 for Implicit Communication as Minimum Entropy Coupling

  Access Paper or Ask Questions

Centralized Model and Exploration Policy for Multi-Agent RL


Jul 14, 2021
Qizhen Zhang, Chris Lu, Animesh Garg, Jakob Foerster

Figure 1 for Centralized Model and Exploration Policy for Multi-Agent RL
Figure 2 for Centralized Model and Exploration Policy for Multi-Agent RL
Figure 3 for Centralized Model and Exploration Policy for Multi-Agent RL
Figure 4 for Centralized Model and Exploration Policy for Multi-Agent RL

  Access Paper or Ask Questions

A New Formalism, Method and Open Issues for Zero-Shot Coordination


Jul 06, 2021
Johannes Treutlein, Michael Dennis, Caspar Oesterheld, Jakob Foerster

Figure 1 for A New Formalism, Method and Open Issues for Zero-Shot Coordination
Figure 2 for A New Formalism, Method and Open Issues for Zero-Shot Coordination
Figure 3 for A New Formalism, Method and Open Issues for Zero-Shot Coordination
Figure 4 for A New Formalism, Method and Open Issues for Zero-Shot Coordination

  Access Paper or Ask Questions

Learned Belief Search: Efficiently Improving Policies in Partially Observable Settings


Jun 16, 2021
Hengyuan Hu, Adam Lerer, Noam Brown, Jakob Foerster

Figure 1 for Learned Belief Search: Efficiently Improving Policies in Partially Observable Settings
Figure 2 for Learned Belief Search: Efficiently Improving Policies in Partially Observable Settings
Figure 3 for Learned Belief Search: Efficiently Improving Policies in Partially Observable Settings
Figure 4 for Learned Belief Search: Efficiently Improving Policies in Partially Observable Settings

  Access Paper or Ask Questions

Quasi-Equivalence Discovery for Zero-Shot Emergent Communication


Mar 14, 2021
Kalesha Bullard, Douwe Kiela, Joelle Pineau, Jakob Foerster

Figure 1 for Quasi-Equivalence Discovery for Zero-Shot Emergent Communication
Figure 2 for Quasi-Equivalence Discovery for Zero-Shot Emergent Communication
Figure 3 for Quasi-Equivalence Discovery for Zero-Shot Emergent Communication
Figure 4 for Quasi-Equivalence Discovery for Zero-Shot Emergent Communication
* 14 pages 
  Access Paper or Ask Questions

Off-Belief Learning


Mar 06, 2021
Hengyuan Hu, Adam Lerer, Brandon Cui, Luis Pineda, David Wu, Noam Brown, Jakob Foerster

Figure 1 for Off-Belief Learning
Figure 2 for Off-Belief Learning
Figure 3 for Off-Belief Learning
Figure 4 for Off-Belief Learning

  Access Paper or Ask Questions

Ridge Rider: Finding Diverse Solutions by Following Eigenvectors of the Hessian


Nov 12, 2020
Jack Parker-Holder, Luke Metz, Cinjon Resnick, Hengyuan Hu, Adam Lerer, Alistair Letcher, Alex Peysakhovich, Aldo Pacchiano, Jakob Foerster

Figure 1 for Ridge Rider: Finding Diverse Solutions by Following Eigenvectors of the Hessian
Figure 2 for Ridge Rider: Finding Diverse Solutions by Following Eigenvectors of the Hessian
Figure 3 for Ridge Rider: Finding Diverse Solutions by Following Eigenvectors of the Hessian
Figure 4 for Ridge Rider: Finding Diverse Solutions by Following Eigenvectors of the Hessian
* Camera-ready version, NeurIPS 2020 
  Access Paper or Ask Questions

Exploring Zero-Shot Emergent Communication in Embodied Multi-Agent Populations


Oct 29, 2020
Kalesha Bullard, Franziska Meier, Douwe Kiela, Joelle Pineau, Jakob Foerster

Figure 1 for Exploring Zero-Shot Emergent Communication in Embodied Multi-Agent Populations
Figure 2 for Exploring Zero-Shot Emergent Communication in Embodied Multi-Agent Populations
Figure 3 for Exploring Zero-Shot Emergent Communication in Embodied Multi-Agent Populations
Figure 4 for Exploring Zero-Shot Emergent Communication in Embodied Multi-Agent Populations

  Access Paper or Ask Questions

The Struggles of Feature-Based Explanations: Shapley Values vs. Minimal Sufficient Subsets


Sep 23, 2020
Oana-Maria Camburu, Eleonora Giunchiglia, Jakob Foerster, Thomas Lukasiewicz, Phil Blunsom

Figure 1 for The Struggles of Feature-Based Explanations: Shapley Values vs. Minimal Sufficient Subsets
Figure 2 for The Struggles of Feature-Based Explanations: Shapley Values vs. Minimal Sufficient Subsets

  Access Paper or Ask Questions

Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning


Mar 19, 2020
Tabish Rashid, Mikayel Samvelyan, Christian Schroeder de Witt, Gregory Farquhar, Jakob Foerster, Shimon Whiteson

Figure 1 for Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning
Figure 2 for Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning
Figure 3 for Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning
Figure 4 for Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning
* Extended version of our ICML 2018 paper (arXiv:1803.11485) 
  Access Paper or Ask Questions

"Other-Play" for Zero-Shot Coordination


Mar 09, 2020
Hengyuan Hu, Adam Lerer, Alex Peysakhovich, Jakob Foerster

Figure 1 for "Other-Play" for Zero-Shot Coordination
Figure 2 for "Other-Play" for Zero-Shot Coordination
Figure 3 for "Other-Play" for Zero-Shot Coordination
Figure 4 for "Other-Play" for Zero-Shot Coordination

  Access Paper or Ask Questions

On the interaction between supervision and self-play in emergent communication


Feb 04, 2020
Ryan Lowe, Abhinav Gupta, Jakob Foerster, Douwe Kiela, Joelle Pineau

Figure 1 for On the interaction between supervision and self-play in emergent communication
Figure 2 for On the interaction between supervision and self-play in emergent communication
Figure 3 for On the interaction between supervision and self-play in emergent communication
Figure 4 for On the interaction between supervision and self-play in emergent communication
* The first two authors contributed equally. Accepted at ICLR 2020 
  Access Paper or Ask Questions

Improving Policies via Search in Cooperative Partially Observable Games


Dec 05, 2019
Adam Lerer, Hengyuan Hu, Jakob Foerster, Noam Brown

Figure 1 for Improving Policies via Search in Cooperative Partially Observable Games
Figure 2 for Improving Policies via Search in Cooperative Partially Observable Games
Figure 3 for Improving Policies via Search in Cooperative Partially Observable Games
Figure 4 for Improving Policies via Search in Cooperative Partially Observable Games
* AAAI 2020 
  Access Paper or Ask Questions

Capacity, Bandwidth, and Compositionality in Emergent Language Learning


Oct 24, 2019
Cinjon Resnick, Abhinav Gupta, Jakob Foerster, Andrew M. Dai, Kyunghyun Cho

Figure 1 for Capacity, Bandwidth, and Compositionality in Emergent Language Learning
Figure 2 for Capacity, Bandwidth, and Compositionality in Emergent Language Learning
Figure 3 for Capacity, Bandwidth, and Compositionality in Emergent Language Learning
Figure 4 for Capacity, Bandwidth, and Compositionality in Emergent Language Learning
* The first two authors contributed equally 
  Access Paper or Ask Questions

Can I Trust the Explainer? Verifying Post-hoc Explanatory Methods


Oct 09, 2019
Oana-Maria Camburu, Eleonora Giunchiglia, Jakob Foerster, Thomas Lukasiewicz, Phil Blunsom

Figure 1 for Can I Trust the Explainer? Verifying Post-hoc Explanatory Methods
Figure 2 for Can I Trust the Explainer? Verifying Post-hoc Explanatory Methods
Figure 3 for Can I Trust the Explainer? Verifying Post-hoc Explanatory Methods
Figure 4 for Can I Trust the Explainer? Verifying Post-hoc Explanatory Methods
* NeurIPS 2019 Workshop Safety and Robustness in Decision Making 
  Access Paper or Ask Questions

Loaded DiCE: Trading off Bias and Variance in Any-Order Score Function Estimators for Reinforcement Learning


Sep 23, 2019
Gregory Farquhar, Shimon Whiteson, Jakob Foerster

Figure 1 for Loaded DiCE: Trading off Bias and Variance in Any-Order Score Function Estimators for Reinforcement Learning
Figure 2 for Loaded DiCE: Trading off Bias and Variance in Any-Order Score Function Estimators for Reinforcement Learning
Figure 3 for Loaded DiCE: Trading off Bias and Variance in Any-Order Score Function Estimators for Reinforcement Learning
Figure 4 for Loaded DiCE: Trading off Bias and Variance in Any-Order Score Function Estimators for Reinforcement Learning

  Access Paper or Ask Questions

A Survey of Reinforcement Learning Informed by Natural Language


Jun 10, 2019
Jelena Luketina, Nantas Nardelli, Gregory Farquhar, Jakob Foerster, Jacob Andreas, Edward Grefenstette, Shimon Whiteson, Tim RocktÀschel

Figure 1 for A Survey of Reinforcement Learning Informed by Natural Language
* Published at IJCAI'19 
  Access Paper or Ask Questions

Differentiable Game Mechanics


May 13, 2019
Alistair Letcher, David Balduzzi, Sebastien Racaniere, James Martens, Jakob Foerster, Karl Tuyls, Thore Graepel

Figure 1 for Differentiable Game Mechanics
Figure 2 for Differentiable Game Mechanics
Figure 3 for Differentiable Game Mechanics
Figure 4 for Differentiable Game Mechanics
* Journal of Machine Learning Research (JMLR), v20 (84) 1-40, 2019 
 * JMLR 2019, journal version of arXiv:1802.05642 
  Access Paper or Ask Questions

On the Pitfalls of Measuring Emergent Communication


Mar 12, 2019
Ryan Lowe, Jakob Foerster, Y-Lan Boureau, Joelle Pineau, Yann Dauphin

Figure 1 for On the Pitfalls of Measuring Emergent Communication
Figure 2 for On the Pitfalls of Measuring Emergent Communication
Figure 3 for On the Pitfalls of Measuring Emergent Communication
Figure 4 for On the Pitfalls of Measuring Emergent Communication
* AAMAS 2019. 13 pages 
  Access Paper or Ask Questions

The StarCraft Multi-Agent Challenge


Feb 26, 2019
Mikayel Samvelyan, Tabish Rashid, Christian Schroeder de Witt, Gregory Farquhar, Nantas Nardelli, Tim G. J. Rudner, Chia-Man Hung, Philip H. S. Torr, Jakob Foerster, Shimon Whiteson

Figure 1 for The StarCraft Multi-Agent Challenge
Figure 2 for The StarCraft Multi-Agent Challenge
Figure 3 for The StarCraft Multi-Agent Challenge
Figure 4 for The StarCraft Multi-Agent Challenge

  Access Paper or Ask Questions

Stable Opponent Shaping in Differentiable Games


Nov 20, 2018
Alistair Letcher, Jakob Foerster, David Balduzzi, Tim RocktÀschel, Shimon Whiteson

Figure 1 for Stable Opponent Shaping in Differentiable Games
Figure 2 for Stable Opponent Shaping in Differentiable Games
Figure 3 for Stable Opponent Shaping in Differentiable Games
Figure 4 for Stable Opponent Shaping in Differentiable Games
* 20 pages, 7 figures 
  Access Paper or Ask Questions

DiCE: The Infinitely Differentiable Monte-Carlo Estimator


Sep 19, 2018
Jakob Foerster, Gregory Farquhar, Maruan Al-Shedivat, Tim RocktÀschel, Eric P. Xing, Shimon Whiteson

Figure 1 for DiCE: The Infinitely Differentiable Monte-Carlo Estimator
Figure 2 for DiCE: The Infinitely Differentiable Monte-Carlo Estimator
Figure 3 for DiCE: The Infinitely Differentiable Monte-Carlo Estimator
Figure 4 for DiCE: The Infinitely Differentiable Monte-Carlo Estimator

  Access Paper or Ask Questions

QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning


Jun 06, 2018
Tabish Rashid, Mikayel Samvelyan, Christian Schroeder de Witt, Gregory Farquhar, Jakob Foerster, Shimon Whiteson

Figure 1 for QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning
Figure 2 for QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning
Figure 3 for QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning
Figure 4 for QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning
* Camera-ready version, International Conference of Machine Learning 2018 
  Access Paper or Ask Questions

The Mechanics of n-Player Differentiable Games


Jun 06, 2018
David Balduzzi, Sebastien Racaniere, James Martens, Jakob Foerster, Karl Tuyls, Thore Graepel

Figure 1 for The Mechanics of n-Player Differentiable Games
Figure 2 for The Mechanics of n-Player Differentiable Games
Figure 3 for The Mechanics of n-Player Differentiable Games
Figure 4 for The Mechanics of n-Player Differentiable Games
* PMLR volume 80, 2018 
 * ICML 2018, final version 
  Access Paper or Ask Questions

Stabilising Experience Replay for Deep Multi-Agent Reinforcement Learning


May 21, 2018
Jakob Foerster, Nantas Nardelli, Gregory Farquhar, Triantafyllos Afouras, Philip H. S. Torr, Pushmeet Kohli, Shimon Whiteson

Figure 1 for Stabilising Experience Replay for Deep Multi-Agent Reinforcement Learning
Figure 2 for Stabilising Experience Replay for Deep Multi-Agent Reinforcement Learning
Figure 3 for Stabilising Experience Replay for Deep Multi-Agent Reinforcement Learning
Figure 4 for Stabilising Experience Replay for Deep Multi-Agent Reinforcement Learning
* Camera-ready version, International Conference of Machine Learning 2017; updated to fix print-breaking image 
  Access Paper or Ask Questions

Counterfactual Multi-Agent Policy Gradients


Dec 14, 2017
Jakob Foerster, Gregory Farquhar, Triantafyllos Afouras, Nantas Nardelli, Shimon Whiteson

Figure 1 for Counterfactual Multi-Agent Policy Gradients
Figure 2 for Counterfactual Multi-Agent Policy Gradients
Figure 3 for Counterfactual Multi-Agent Policy Gradients
Figure 4 for Counterfactual Multi-Agent Policy Gradients

  Access Paper or Ask Questions