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

Add to Chrome

Add to Firefox

Picture for Philip J. Ball

Philip J. Ball

Learning General World Models in a Handful of Reward-Free Deployments

Oct 23, 2022
Yingchen Xu, Jack Parker-Holder, Aldo Pacchiano, Philip J. Ball, Oleh Rybkin, Stephen J. Roberts, Tim Rocktäschel, Edward Grefenstette

Figure 1 for Learning General World Models in a Handful of Reward-Free Deployments

Figure 2 for Learning General World Models in a Handful of Reward-Free Deployments

Figure 3 for Learning General World Models in a Handful of Reward-Free Deployments

Figure 4 for Learning General World Models in a Handful of Reward-Free Deployments

* To be published at NeurIPS 2022. Code and videos available at https://ycxuyingchen.github.io/cascade/

Access Paper or Ask Questions

Bayesian Generational Population-Based Training

Jul 19, 2022
Xingchen Wan, Cong Lu, Jack Parker-Holder, Philip J. Ball, Vu Nguyen, Binxin Ru, Michael A. Osborne

Figure 1 for Bayesian Generational Population-Based Training

Figure 2 for Bayesian Generational Population-Based Training

Figure 3 for Bayesian Generational Population-Based Training

Figure 4 for Bayesian Generational Population-Based Training

* AutoML Conference 2022. 10 pages, 4 figure, 3 tables (28 pages, 10 figures, 7 tables including references and appendices)

Access Paper or Ask Questions

Stabilizing Off-Policy Deep Reinforcement Learning from Pixels

Jul 03, 2022
Edoardo Cetin, Philip J. Ball, Steve Roberts, Oya Celiktutan

Figure 1 for Stabilizing Off-Policy Deep Reinforcement Learning from Pixels

Figure 2 for Stabilizing Off-Policy Deep Reinforcement Learning from Pixels

Figure 3 for Stabilizing Off-Policy Deep Reinforcement Learning from Pixels

Figure 4 for Stabilizing Off-Policy Deep Reinforcement Learning from Pixels

* Short presentation at ICML 2022

Access Paper or Ask Questions

Revisiting Design Choices in Model-Based Offline Reinforcement Learning

Oct 08, 2021
Cong Lu, Philip J. Ball, Jack Parker-Holder, Michael A. Osborne, Stephen J. Roberts

Figure 1 for Revisiting Design Choices in Model-Based Offline Reinforcement Learning

Figure 2 for Revisiting Design Choices in Model-Based Offline Reinforcement Learning

Figure 3 for Revisiting Design Choices in Model-Based Offline Reinforcement Learning

Figure 4 for Revisiting Design Choices in Model-Based Offline Reinforcement Learning

* Spotlight @ RL4RealLife Workshop ICML2021

Access Paper or Ask Questions

Augmented World Models Facilitate Zero-Shot Dynamics Generalization From a Single Offline Environment

Apr 12, 2021
Philip J. Ball, Cong Lu, Jack Parker-Holder, Stephen Roberts

Figure 1 for Augmented World Models Facilitate Zero-Shot Dynamics Generalization From a Single Offline Environment

Figure 2 for Augmented World Models Facilitate Zero-Shot Dynamics Generalization From a Single Offline Environment

Figure 3 for Augmented World Models Facilitate Zero-Shot Dynamics Generalization From a Single Offline Environment

Figure 4 for Augmented World Models Facilitate Zero-Shot Dynamics Generalization From a Single Offline Environment

* To be presented as a Spotlight at the "Self-Supervision for Reinforcement Learning Workshop" @ ICLR 2021

Access Paper or Ask Questions

OffCon$^3$: What is state of the art anyway?

Jan 27, 2021
Philip J. Ball, Stephen J. Roberts

Figure 1 for OffCon$^3$: What is state of the art anyway?

Figure 2 for OffCon$^3$: What is state of the art anyway?

Figure 3 for OffCon$^3$: What is state of the art anyway?

Figure 4 for OffCon$^3$: What is state of the art anyway?

Access Paper or Ask Questions