研究人员开发了新的强化学习策略方法，旨在提高效率和表达能力。一种方法，基于分数的一次性均值流策略优化（SOM），使用Q函数分数和概率流ODE构建目标速度场，通过减少训练和推理时间，在在线强化学习中实现了最先进的性能。另一项开发，随机均值流策略（SMFP），提供了一个一次性生成策略类别，通过均值流变换将噪声映射到动作，为离策略设置中稳定和探索性的策略改进提供了统一的目标。 AI
影响 这些新的策略优化技术有望加快强化学习的训练和推理速度，从而可能加速机器人和自主系统的进步。
排序理由 该集群包含两篇详细介绍强化学习新方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 6 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
arXiv:2605.23365v1 Announce Type: cross Abstract: Diffusion and flow matching have emerged as expressive policy classes in reinforcement learning, but their reliance on multi-step denoising imposes substantial computational overhead at inference time, which is particularly proble…
Diffusion and flow matching have emerged as expressive policy classes in reinforcement learning, but their reliance on multi-step denoising imposes substantial computational overhead at inference time, which is particularly problematic in online RL. MeanFlow offers a promising al…
Diffusion and flow matching have emerged as expressive policy classes in reinforcement learning, but their reliance on multi-step denoising imposes substantial computational overhead at inference time, which is particularly problematic in online RL. MeanFlow offers a promising al…
arXiv:2605.21282v2 Announce Type: cross Abstract: Online off-policy reinforcement learning (RL) is shaped by two coupled choices: the policy class and the update rule. Gaussian policies are fast and have tractable entropy, but struggle with multimodal action distributions. Genera…
Online off-policy reinforcement learning (RL) is shaped by two coupled choices: the policy class and the update rule. Gaussian policies are fast and have tractable entropy, but struggle with multimodal action distributions. Generative policies are more expressive, but often requi…
Online off-policy reinforcement learning (RL) is shaped by two coupled choices: the policy class and the update rule. Gaussian policies are fast and have tractable entropy, but struggle with multimodal action distributions. Generative policies are more expressive, but often requi…