arXiv cs.AI· Yuanchu Liang, Yue Yang, An-Chi He, Hanna Kurniawati·· 12 小时前AI 评分31
LARBRID：用潜在噪声表示桥接实现高效扩散策略微调
Efficient Diffusion Policy Fine Tuning with Latent Noise Representation Bridging
AI 导读
针对预训练扩散与流策略在强化学习微调中计算昂贵、样本效率低的问题，论文提出 Latent Noise Linear Markov Decision Process（LNL-MDP）并推导 Bridge Equation，通过让噪声空间与动作空间共享表示来消除"噪声混叠"。
正文 · 原文
Abstract:Pre-trained diffusion and flow policies have emerged as powerful backbones for robotics, yet adapting them to novel data via reinforcement learning (RL) remains computationally expensive, sample inefficient and lacks theoretical guarantees. Prior approaches that fine-tune the initial noise bypass costly backpropagation through time but struggle with the "noise aliasing" problem: there does not exista natural link between the noise space value and the action space value. This paper resolves these challenges by introducing the Latent Noise Linear Markov Decision Process (LNL-MDP). We derive the Bridge Equation for LNL-MDPs which eliminates the noise aliasing problem by aligning values with a shared representation between the noise and the action space. Furthermore, our LNL-MDP fine tuning framework falls under the FineTuneRL theory with near-optimal guarantee. Leveraging these insights, we propose Latent Representation Bridging for Diffusion (LARBRID), an algorithm that efficiently learns the representation to optimize the initial distribution that steers the pre-trained model to act optimally in the environment. It yields up to 2-3$\times$ performance increases on challenging control problems across a diverse suite of robotic locomotion and manipulation tasks. Code can be found at: this http URL.
| Comments: | Conference on Neural Information Processing Systems (NeurIPS), 2026 |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.07275 [cs.RO] |
| (or arXiv:2611.07275v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2611.07275 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yuanchu Liang [view email]
[v1]
Thu, 5 Nov 2026 00:49:53 UTC (4,322 KB)
来源：arXiv cs.AI · arxiv.org