CausalWAM separates action causes from scene prediction in world models
Original title: CausalWAM: Learning the Causes of Actions in World Action Models
CausalWAM adds sparse routing to pretrained WAMs proving action-relevant latents are identifiable up to permutation for improved robotic closed-loop control.
World action models use latent variables for scene evolution and action generation. The method separates action causes from nuisance variables under sparsity constraints. It proves the latents for actions are invertible functions of action-relevant ones without needing temporal dependence assumptions. Experiments on synthetic data and robots validate the approach and show efficiency gains.
- CausalWAM incorporates sparsity into pretrained WAM via sparse routing path
- Action-relevant latents identified up to permutation under sparsity
- No sufficient temporal dependence or regime evolution assumptions required
- Improved closed-loop success on real-world robots
- AI models
- CausalWAM
From the original article
1 min read at huggingface.co
World action models (WAMs) learn latent variables from observations and use these to both predict the evolution of the scene and to generate actions. The prediction has to be of the whole scene, whereas only a portion of the scene causes actions.
By Juntao Cheng, Jinzhou Tang, Fan Feng · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.