{"slug": "reinforcement-learning-via-fixed-point-flow-matching", "title": "Reinforcement Learning via Fixed-point Flow Matching", "summary": "Researchers introduced fixed-point flow matching (FFM), a data-free procedure for training and fine-tuning flow policies in reinforcement learning, detailed in arXiv paper 2610.12848v1. FFM formulates the actor update as a fixed-point iteration using a flow-matching-style regression objective in which the critic's action gradients serve as regression targets, and the authors show the optimal policy is a fixed point of this iteration. Experiments showed FFM fine-tunes pretrained flow policies in the offline-to-online setting and trains flow policies from scratch in online RL without requiring training data.", "body_md": "arXiv:2610.12848v1 Announce Type: new \nAbstract: In maximum entropy and behavior-constrained reinforcement learning (RL), the optimal policy is given by a Boltzmann distribution. Generating samples from this distribution can be challenging, as it is often complex and multimodal. While flow policies trained with flow matching provide a powerful framework for representing such distributions, they traditionally rely on access to training data, which is generally unavailable in online RL. To address this limitation, we propose fixed-point flow matching (FFM), a data-free procedure for training and fine-tuning flow policies in RL. FFM leverages the target flow identity to formulate the actor update as a fixed-point iteration, utilizing a flow-matching-style regression objective where the critic's action gradients serve as regression targets. We show that the optimal policy is a fixed point of this iteration. Through extensive experiments, we demonstrate that FFM effectively fine-tunes pretrained flow policies in the offline-to-online setting and enables efficient and accurate training of flow policies from scratch in online RL. These results establish fixed-point flow matching as a principled approach for learning expressive flow-based policies without requiring training data.", "url": "https://wpnews.pro/news/reinforcement-learning-via-fixed-point-flow-matching", "canonical_source": "https://www.machinebrief.com/news/reinforcement-learning-via-fixed-point-flow-matching-5f0y", "published_at": "2026-10-12 04:00:00+00:00", "updated_at": "2026-10-12 06:07:50.032652+00:00", "lang": "en", "topics": ["machine-learning", "artificial-intelligence", "ai-research", "neural-networks"], "entities": ["arXiv", "fixed-point flow matching", "FFM"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/reinforcement-learning-via-fixed-point-flow-matching", "markdown": "https://wpnews.pro/news/reinforcement-learning-via-fixed-point-flow-matching.md", "text": "https://wpnews.pro/news/reinforcement-learning-via-fixed-point-flow-matching.txt", "jsonld": "https://wpnews.pro/news/reinforcement-learning-via-fixed-point-flow-matching.jsonld"}}