<ul><li>Dreamweaver is a neural architecture designed to discover hierarchical and compositional representations from raw videos.</li><li>The model leverages a Recurrent Block-Slot Unit (RBSU) to decompose videos into their constituent objects and attributes.</li><li>Dreamweaver outperforms current state-of-the-art baselines for world modeling and allows the generation of novel videos by recombining attributes from previously seen objects.</li><li>The research is evaluated under the DCI framework across multiple datasets.</li></ul>

Dreamweaver: Learning Compositional World Models from Pixels

Discover more