arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.01965cs.LGmath.OC
收藏

通过凸优化分析基于神经网络的生成扩散模型

Analyzing Neural Network-Based Generative Diffusion Models through Convex Optimization

Stanford University(斯坦福大学)

Fangzhao Zhang, Mert Pilanci

AI总结:

本研究通过将分数匹配和去噪分数匹配转化为凸优化,为基于两层神经网络的扩散模型建立了理论框架,证明了浅层网络训练可归结为单个凸程序,并在有限数据下精确刻画了预测分数函数与收敛性。

摘要:

扩散模型在尖端的图像、视频和音频生成中正获得广泛使用。基于分数的扩散模型在这些方法中脱颖而出,需要估计输入数据分布的分数函数。在本研究中,我们提出了一个理论框架,通过将分数匹配和去噪分数匹配重新表述为凸优化问题,来分析基于两层神经网络的扩散模型。我们证明,训练用于分数预测的浅层神经网络可以通过求解单个凸程序来完成。尽管大多数扩散模型的分析是在渐近设置下进行或依赖近似,我们刻画了精确的预测分数函数,并为具有有限数据的基于神经网络的扩散模型建立了收敛结果。我们的结果提供了对基于神经网络的扩散模型在非渐近设置中所学习内容的精确刻画。

英文摘要:

Diffusion models are gaining widespread use in cutting-edge image, video, and audio generation. Score-based diffusion models stand out among these methods, necessitating the estimation of score function of the input data distribution. In this study, we present a theoretical framework to analyze two-layer neural network-based diffusion models by reframing score matching and denoising score matching as convex optimization. We prove that training shallow neural networks for score prediction can be done by solving a single convex program. Although most analyses of diffusion models operate in the asymptotic setting or rely on approximations, we characterize the exact predicted score function and establish convergence results for neural network-based diffusion models with finite data. Our results provide a precise characterization of what neural network-based diffusion models learn in non-asymptotic settings.

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。

↑