Scholarlink

GPs provide an alternative approach to the regression just addressed [1, 2]. If we consider the linear regression \(Y=\theta _0+\theta _1 X\) , Bayesian linear regression provides a probabilistic way to estimate these coefficients from the collected data, within a parametric framework. On the contrary GP represents a non-parametric approach, finding over all possible functions f(X) those consistent with the available data, that is, something like we have an infinity of parameters. Like all the Bayesian methods, it starts with some priors that are updated from the data to avoid considering all possible functions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Radom Variables: Gaussian Processes, GP

  • 章节

  • 出版模式：

    开放获取

  • 发表日期：

    2025年

查看全文
  • Francisco Chinesta,
  • Elías Cueto,
  • Victor Champaney,
  • Chady Ghnatios,
  • Amine Ammar,
  • Nicolas Hascoët,
  • David González,
  • Icíar Alfaro,
  • Daniele Di Lorenzo,
  • Angelo Pasquale,
  • Dominique Baillargeat

摘要

GPs provide an alternative approach to the regression just addressed [1, 2]. If we consider the linear regression \(Y=\theta _0+\theta _1 X\) , Bayesian linear regression provides a probabilistic way to estimate these coefficients from the collected data, within a parametric framework. On the contrary GP represents a non-parametric approach, finding over all possible functions f(X) those consistent with the available data, that is, something like we have an infinity of parameters. Like all the Bayesian methods, it starts with some priors that are updated from the data to avoid considering all possible functions.