{"ID":2830863,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2512.10016","arxiv_id":"2512.10016","title":"Latent Action World Models for Control with Unlabeled Trajectories","abstract":"Inspired by how humans combine direct interaction with action-free experience (e.g., videos), we study world models that learn from heterogeneous data. Standard world models typically rely on action-conditioned trajectories, which limits effectiveness when action labels are scarce. We introduce a family of latent-action world models that jointly use action-conditioned and action-free data by learning a shared latent action representation. This latent space aligns observed control signals with actions inferred from passive observations, enabling a single dynamics model to train on large-scale unlabeled trajectories while requiring only a small set of action-labeled ones. We use the latent-action world model to learn a latent-action policy through offline reinforcement learning (RL), thereby bridging two traditionally separate domains: offline RL, which typically relies on action-conditioned data, and action-free training, which is rarely used with subsequent RL. On the DeepMind Control Suite, our approach achieves strong performance while using about an order of magnitude fewer action-labeled samples than purely action-conditioned baselines. These results show that latent actions enable training on both passive and interactive data, which makes world models learn more efficiently.","short_abstract":"Inspired by how humans combine direct interaction with action-free experience (e.g., videos), we study world models that learn from heterogeneous data. Standard world models typically rely on action-conditioned trajectories, which limits effectiveness when action labels are scarce. We introduce a family of latent-actio...","url_abs":"https://arxiv.org/abs/2512.10016","url_pdf":"https://arxiv.org/pdf/2512.10016v1","authors":"[\"Marvin Alles\",\"Xingyuan Zhang\",\"Patrick van der Smagt\",\"Philip Becker-Ehmck\"]","published":"2025-12-10T19:09:45Z","proceeding":"cs.LG","tasks":"[\"cs.LG\"]","methods":"[\"Reinforcement Learning\"]","has_code":false}