{"slug": "flow-jepa-flow-matching-for-robust-latent-dynamics-in-jepa-world-models", "title": "Flow-JEPA: Flow Matching for Robust Latent Dynamics in JEPA World Models", "summary": "Researchers propose Flow-JEPA (F-JEPA), a conditional flow matching dynamics model for JEPA world models that jointly generates future latent states, improving mean success from 86% to 92% under clean observations and from 67% to 86% under noisy conditions compared to the deterministic LeWorldModel baseline.", "body_md": "arXiv:2608.29029v1 Announce Type: new\nAbstract: Joint-Embedding Predictive Architectures (JEPAs) have shown strong potential for learning compact predictive representations, and LeWorldModel (LeWM) extends this paradigm to reconstruction-free latent world modeling from pixels. However, its deterministic autoregressive predictor generates future states through repeated one-step transitions, which can accumulate errors and remain sensitive to task-irrelevant visual perturbations. In this work, we propose Flow-JEPA (F-JEPA), a conditional flow matching dynamics model that jointly generates a sequence of future latent states conditioned on the current observation and actions. A Gaussian distribution serves as the flow source, exposing the vector field to perturbed latent trajectories as it learns to transport them toward clean future representations. This formulation retains the reconstruction-free JEPA framework while replacing point-wise transition regression with stochastic trajectory-level prediction. F-JEPA raises mean success from $86\\%$ to $92\\%$ under clean observations and from $67\\%$ to $86\\%$ under noisy conditions, suggesting that conditional flow matching provides a promising alternative to deterministic autoregressive dynamics in JEPA world models.", "url": "https://wpnews.pro/news/flow-jepa-flow-matching-for-robust-latent-dynamics-in-jepa-world-models", "canonical_source": "https://arxiv.org/abs/2608.29029", "published_at": "2026-09-01 04:00:00+00:00", "updated_at": "2026-09-01 04:25:55.650268+00:00", "lang": "en", "topics": ["machine-learning", "artificial-intelligence", "ai-research"], "entities": ["Flow-JEPA", "LeWorldModel", "Joint-Embedding Predictive Architectures"], "alternates": {"html": "https://wpnews.pro/news/flow-jepa-flow-matching-for-robust-latent-dynamics-in-jepa-world-models", "markdown": "https://wpnews.pro/news/flow-jepa-flow-matching-for-robust-latent-dynamics-in-jepa-world-models.md", "text": "https://wpnews.pro/news/flow-jepa-flow-matching-for-robust-latent-dynamics-in-jepa-world-models.txt", "jsonld": "https://wpnews.pro/news/flow-jepa-flow-matching-for-robust-latent-dynamics-in-jepa-world-models.jsonld"}}