back

Learning Discrete Latent Dynamics for Planning

Investigating finite-state representations of a world model and learning optimal policies via policy iteration over the learned dynamics.

Open in Google Slides