arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2310.04490cs.LGphysics.class-ph
收藏

从作用量原理出发的生成扩散

Generative Diffusion From An Action Principle

Kavli Institute for Cosmological Physics(卡夫利宇宙物理研究所);University of Chicago(芝加哥大学)

Akhil Premkumar

AI总结:

本文从作用量原理推导得分匹配,将逆向扩散视为最优控制问题,揭示不同扩散模型间的联系。

摘要:

生成扩散模型通过逆转一个将给定数据集转换为通用噪声的扩散过程来合成新样本。这通过训练神经网络匹配给定数据集概率分布对数的梯度(也称为得分)来实现。将逆向扩散视为一个最优控制问题,我们表明得分匹配可以从作用量原理推导出来,类似于物理学中常用的原理。我们利用这一见解来展示不同类别扩散模型之间的联系。

英文摘要:

Generative diffusion models synthesize new samples by reversing a diffusive process that converts a given data set to generic noise. This is accomplished by training a neural network to match the gradient of the log of the probability distribution of a given data set, also called the score. By casting reverse diffusion as an optimal control problem, we show that score matching can be derived from an action principle, like the ones commonly used in physics. We use this insight to demonstrate the connection between different classes of diffusion models.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑