Model基于得分的生成模型

Score-Based Generative Model

A generative modeling paradigm based on score matching that generates samples by estimating the score function of the data distribution and combining it with stochastic differential equations for sampling.

Timeline (last 90 days)

Oct 16

Classical diffusion models (such as DDPM and Score-Based Generative Model) rely on Brownian motion driven by Gaussian noise as the forward corruption process.

Unverified50%

All Facts (1)

Unverified

Classical diffusion models (such as DDPM and Score-Based Generative Model) rely on Brownian motion driven by Gaussian noise as the forward corruption process.

50%

Source Articles

StochLOB: Heavy-Tailed Score Matching Improves the Robustness of Order Book Trend PredictionOct 16