Sie sind hier: Startseite › Teaching › SS2025 › World Models (Seminar)

World Models (Seminar)

 Course information

 

 

Overview

In recent years, world models have emerged as a powerful paradigm in machine learning, particularly in the realms of reinforcement learning, robotics, and generative modeling. Inspired by how humans learn internal representations of their environment, world models aim to build compact, predictive models of the world that agents can use to plan, simulate, and make decisions.

This seminar will explore the foundations, architectures, and applications of world models, focusing on how they enable agents to "dream" about future states, and ultimately facilitate the learning process.

 

Format

The course will be given in person, in the form of a block seminar, where papers are read and presented by students in the form of a scientific poster.

 

Application and Paper Voting Process

Application process

Due to the unexpectedly high number of applications, we kindly ask all students who are seriously interested in attending to register for the course as their first priority. Additionally, we ask you to submit a short motivation statement explaining why you would like to join the seminar and the necessity for the seminar credits. 

Paper Voting Procedure

We have released the paper list on ILIAS and the lab's webpage (https://nr.informatik.uni-freiburg.de/teaching/ss2025/world-models). Please review the paper list and select three papers that interest you the most. Based on your choices, we will try to find a fair assignment. Please send your preference (for example: 3>10>1) and your matriculation number to rlexam@tf.uni-freiburg.de by 18:00 June 13th.

 

Timeline

Date Done Comment
Introductory Lecture Tuesday, April 22nd ✓ This will take place at 10:00 am - 12:00 pm, Nexus-Lab (1st Floor) @ IMBIT, Georges-Köhler-Allee 201
Motivation Mail Monday, April 28th ✓

Please send your motivation (a few sentences is sufficient) to rlexam@tf.uni-freiburg.de by 6:00 p.m. April 28th.

Announcement Paper List Monday, June 9th ✓ Papers are listed below.
Paper Voting Deadline Friday, June 13th  ✓ Please send your paper preferences to rlexam@tf.uni-freiburg.de by 6:00 p.m. June 12th. 
First Meeting with Supervisor before Monday, July 7th   Your advisor will contact you after the papers are assigned to let you know which paper you will be working on and to schedule some meetings with you.

This is your opportunity to ask questions about the content of the paper.
Second Meeting with Supervisor before Monday, July 21st This is your opportunity to go over the poster draft with your supervisor.
Final Poster Deadline

Monday, July 28th

   Please send a pdf of the poster to rlexam@tf.uni-freiburg.de by 6:00 p.m. July 28th. 
Block Seminar Friday, August 1st During the final block seminar, you will present your poster to the supervisors and your fellow students attending the seminar.

 

Paper List

Index Paper Link
1 Reasoning with Language Model is Planning with World Model https://arxiv.org/abs/2305.14992
2 AdaWorld: Learning Adaptable World Models with Latent Actions https://arxiv.org/abs/2503.18938
3 Planning to Explore via Self-Supervised World Models http://arxiv.org/abs/2005.05960
4 Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture https://arxiv.org/abs/2301.08243
5

Learning and Leveraging World Models in Visual
Representation Learning

https://arxiv.org/abs/2403.00504
6 Temporal Difference Learning for Model Predictive Control https://arxiv.org/abs/2203.04955
7 Dream to Drive with Predictive Individual World Model https://arxiv.org/pdf/2501.16733
8 M3PC: Test-time Model Predictive Control for Pretrained Masked Trajectory Model https://arxiv.org/html/2412.05675v1
9 Genie: Generative Interactive Environments https://arxiv.org/abs/2402.15391
10 Compete and Compose: Learning Independent Mechanisms for Modular World Models https://arxiv.org/pdf/2404.15109
11 Multimodal foundation world models for generalist embodied agents https://arxiv.org/abs/2406.18043
12 World Models http://arxiv.org/abs/1803.10122
13 Latent Linear Quadratic Regulator for Robotic Control Tasks https://arxiv.org/abs/2407.11107v2
14 A shared robot control system combining augmented reality and motor imagery brain–computer interfaces with eye tracking https://iopscience.iop.org/article/10.1088/1741-2552/ad7f8d/pdf
15 Accelerating Model-Based Reinforcement Learning with State-Space World Models https://arxiv.org/abs/2502.20168
16 Dreaming of Many Worlds: Learning Contextual World Models aids Zero-Shot Generalization https://openreview.net/forum?id=o8DrRuBsQb
17 Revisiting Feature Prediction for Learning Visual Representations from Video https://ai.meta.com/research/publications/revisiting-feature-prediction-for-learning-visual-representations-from-video/

 

 

 

Resources

Poster guideline as pptx or pdf