1:30 PM
Oumayma Bounou
New York University
Planning with world models: control-aware representations and gradient-based optimization
SPECIAL EVENT · OCTOBER 29, 2026
Mini-Workshop on World Models
An afternoon of research conversations on world models for prediction, planning, control, reinforcement learning, and robotics, bringing together speakers from Columbia University, NYU, and USC.
THURSDAY · OCTOBER 29
The program
Armen Avanessians Conference Room · 14th Floor (DSI Suite), Northwest Corner Building
Planning & control
2:00 PM
Stephen Tu
University of Southern California
Short-Context Prediction in Continuous-State Dynamical Systems
2:30 PM
Break
3:00 PM
Adam Block
Columbia University
Improving Exploration in RLVR through Retries
3:30 PM
Yunzhu Li
Columbia University · Remote talk
Structured World Models as Scalable Data Engines for Robotics
4:00 PM
Break
4:15 PM
Rohun Agrawal
Columbia University
Closing the Train-Test Gap in World Models for Gradient-based Planning
4:45 PM
Ying Wang
New York University
Towards World Models That Learn and Adapt in an Open World
5:15 PM
Group Discussion
Open discussion with workshop participants
THE SPEAKERS
Speakers
Researchers and their talks from the October 29 mini-workshop.
OB
Oumayma Bounou
New York University
Control-aware representations and planningOumayma Bounou is a postdoctoral researcher at NYU working with Yann LeCun on world models. She earned her PhD through the WILLOW team at Inria and École Normale Supérieure, advised by Justin Carpentier and Jean Ponce. Her research spans world models, control, and optimization.
Research profile ↗ST
Stephen Tu
University of Southern California
Short-context prediction in dynamical systemsStephen Tu is an Assistant Professor of Electrical and Computer Engineering at USC, with a courtesy appointment in Computer Science. His research connects statistical learning, dynamical systems, and control, with applications to robotics. Previously, he was a research scientist on the Google DeepMind robotics team.
Research profile ↗AB
Adam Block
Columbia University
Exploration in reinforcement learningAdam Block is an Assistant Professor of Computer Science at Columbia University, with an affiliation in Electrical Engineering. He studies the theoretical foundations of machine learning, especially efficient algorithms for reinforcement learning, sequential decision-making, and language models. Before Columbia, he completed his PhD at MIT and a postdoc at Microsoft Research.
Research profile ↗YL
Yunzhu Li
Columbia University · Remote
Structured world models for roboticsYunzhu Li is an Assistant Professor of Computer Science at Columbia University, where he leads research in robotic perception, interaction, and learning. His work combines robotics, computer vision, and machine learning to help robots understand and interact with the physical world. He received his PhD from MIT.
Research profile ↗RA
Rohun Agrawal
Columbia University
Gradient-based planning with world modelsRohun Agrawal is a Computer Science PhD student at Columbia University, co-advised by Micah Goldblum and Pavel Izmailov, and a visiting researcher at NYU CILVR. His work aims to extend the time horizons of AI agents through world-model planning and memory retrieval. He previously studied applied and computational mathematics at Caltech.
Research profile ↗YW
Ying Wang
New York University
Learning and adaptation in an open worldYing Wang is a PhD student at NYU's Center for Data Science, advised by Mengye Ren and Yann LeCun. Her research focuses on world models, multimodal learning, reasoning, and planning. She holds an MS in Data Science from NYU and an undergraduate degree from McGill University.
Research profile ↗MOMENTS FROM THE WORKSHOP
A photo story
A visual record of the talks, research exchanges, and conversations around the workshop. Photographs appear here as they are added to the archive.
Photo story coming soon
Photographs from the talks and group discussion will appear here as they are added to the workshop archive.
This event was kindly sponsored by the Data Science Institute Center for Physical AI at Columbia University.