Saved in:
Bibliographic Details
Main Authors: Hauri, Michael, Zenke, Friedemann
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.07083
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • Model-based reinforcement learning (MBRL) agents operating in high-dimensional observation spaces, such as Dreamer, rely on learning abstract representations for effective planning and control. Existing approaches typically employ reconstruction-based objectives in the observation space, which can render representations sensitive to task-irrelevant details. Recent alternatives trade reconstruction for auxiliary action prediction heads or view augmentation strategies, but perform worse in the Crafter environment than reconstruction-based methods. We close this gap between Dreamer and reconstruction-free models by introducing a JEPA-style predictor defined on continuous, deterministic representations. Our method matches Dreamer's performance on Crafter, demonstrating effective world model learning on this benchmark without reconstruction objectives.