Schedule
08:15 - 08:45 Poster Session I.
08:45 - 09:00 Opening Remarks.
09:00 - 09:30 Keynote # 1 Xiaolong Wang & Nicklas Hansen (UCSD): World Models on a Budget.
09:30 - 10:00 Keynote # 2 Chelsea Finn (Stanford & Physical Intelligence): Developing Generalist Vision-Language-Action Models.
10:00 - 10:15 Coffee Break
10:15 - 10:30 Industry Demo Anthony Hu (Wayve): GAIA-2: A Generative World Model for Autonomous Driving.
10:30 - 11:00 Keynote # 3 Tim Rocktäschel (UCL & Google DeepMind) & Jack Parker-Holder (Google DeepMind): Scaling Foundation World Models.
11:00 - 11:30 Keynote # 4 Furong Huang (University of Maryland): Foundation Model for Sequential Decision-Making: A World-Model Learning Perspective.
11:30 - 12:00 Keynote # 5 Jeff Clune (UBC): Open-ended Agent Learning in the Era of Foundation Models and Foundation World Models.
12:00 - 13:00 Lunch Break + Poster Session II
13:00 - 13:30 Keynote # 6 Hong Zhou: Hypotheses & Visions for an Intelligent World.
13:30 - 14:00 Keynote # 7 Stefano Ermon (Stanford University): Diffusion Language Models: Towards A Unifying Paradigm for Multimodal Generative Modeling.
14:00 - 14:50 Panel Discussion: Current Development and Future Challenges of World Models.
Moderator: Tim Rocktäschel (UCL & Google DeepMind), Special Guest - Jürgen Schmidhuber (KAUST & IDSIA).
Panelists: Jeff Clune (UBC), Stefano Ermon (Stanford), Kun Zhang (CMU), Furong Huang (UMD), David Ha (Sakana.AI), Elahe Arani (Wayve)
14:50 - 15:30 Coffee Break + Poster Session III
15:30 - 16:00 Oral Presentation I
15:30 - 15:40 Improving Transformer World Models for Data-Efficient RL. Joseph Ortiz, Jyothi Swaroop Guntupalli, Kevin Murphy (Google DeepMind)
15:40 - 15:50 From Foresight to Forethought: VLM-in-the-Loop Policy Steering via Latent Alignment. Yilin Wu, Michelle Zhao (CMU)
15:50 - 16:00 When do neural networks learn world models? Tianren Zhang (Tsinghua University)
16:00 - 16:30 Keynote # 8 Jakob Foerster (Oxford University): The Simulation Hypothesis: Opportunities and Challenges.
16:30 - 17:00 Keynote # 9 Tom Everitt (Google DeepMind): Robust Agents Learn Causal World Models.
17:00 - 17:30 Oral Presentation II
17:00 - 17:10 Scalable Humanoid Whole-Body Control via Differentiable Neural Network Dynamics. Yu Lei (SJTU)
17:10 - 17:20 Masked Generative Priors Improve World Models Sequence Modelling Capabilities. Cristian Meo (Delft University of Technology), Zarif Ikram (NUS)
17:20 - 17:30 Temporal Difference Flows. Jesse Farebrother (Meta AI & McGill University)
17:30 - 18:30 Paper Award & Closing Remarks & Social
Invited Speakers & Panelist
UBC & CIFAR AI
Stanford University
UCL & Google DeepMind
University of Oxford
Stanford University & Physical Intelligence
CMU & MBZUAI
University of Maryland
UCSD