Peter Stone's Selected Publications

Classified by TopicClassified by Publication TypeSorted by DateSorted by First Author Last NameClassified by Funding Source


Factored Latent Action World Models

Factored Latent Action World Models.
Zizhao Wang, Chang Shi, Jiaheng Hu, Kevin Rohling, Roberto Martín-Martín, Amy Zhang, and Peter Stone .
In International Conference on Machine Learning, July 2026.

Download

[PDF]6.3MB  

Abstract

Learning latent actions from action-free video has emerged as a powerful paradigm for scaling up controllable world model learning. Latent actions provide a natural interface for users to iteratively generate and manipulate videos. However, most existing approaches rely on monolithic inverse and forward dynamics models that learn a single latent action to control the entire scene, and therefore struggle in complex environments where multiple entities act simultaneously. This paper introduces Factored Latent Action Model (FLAM), a factored dynamics framework that decomposes the scene into independent factors, each inferring its own latent action and predicting its own next-step factor value. This factorized structure enables more accurate modeling of complex multi-entity dynamics and improves video generation quality in action-free video settings compared to monolithic models. Based on experiments on both simulation and real-world multi-entity datasets, we find that FLAM outperforms prior work in prediction accuracy and representation quality, and facilitates downstream policy learning, demonstrating the benefits of factorized latent action models.

BibTeX Entry

@InProceedings{zzwang_flam,
  author   = {Zizhao Wang and Chang Shi and Jiaheng Hu and Kevin Rohling and Roberto Martín-Martín and Amy Zhang and Peter Stone },
  title    = {Factored Latent Action World Models},
  booktitle = {International Conference on Machine Learning},
  year     = {2026},
  month    = {July},
  location = {Seoul, South Korea},
  abstract = {Learning latent actions from action-free video has emerged as a powerful paradigm for scaling up controllable world model learning. Latent actions provide a natural interface for users to iteratively generate and manipulate videos. However, most existing approaches rely on monolithic inverse and forward dynamics models that learn a single latent action to control the entire scene, and therefore struggle in complex environments where multiple entities act simultaneously. This paper introduces Factored Latent Action Model (FLAM), a factored dynamics framework that decomposes the scene into independent factors, each inferring its own latent action and predicting its own next-step factor value. This factorized structure enables more accurate modeling of complex multi-entity dynamics and improves video generation quality in action-free video settings compared to monolithic models. Based on experiments on both simulation and real-world multi-entity datasets, we find that FLAM outperforms prior work in prediction accuracy and representation quality, and facilitates downstream policy learning, demonstrating the benefits of factorized latent action models.},
}

Generated by bib2html.pl (written by Patrick Riley ) on Wed Jun 10, 2026 15:26:44