Scholarlink

This chapter presents causal discovery from temporal data. It starts by introducing dynamic Bayesian networks, as a way to represent a causal dynamic structure. The main focus is on multivariate time-series, describing several techniques for causal discovering, including Granger causality, constraint-based and score-based approaches, and linear models with non-Gaussian noise. Then it analyzes causal discovery from event sequences. It finalizes by introducing the problem of subsampling, and different strategies to cope with this problem.

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

Causal Discovery from Temporal Data

  • 章节

  • 发表日期：

    2026年

查看全文
  • Luis Enrique Sucar

摘要

This chapter presents causal discovery from temporal data. It starts by introducing dynamic Bayesian networks, as a way to represent a causal dynamic structure. The main focus is on multivariate time-series, describing several techniques for causal discovering, including Granger causality, constraint-based and score-based approaches, and linear models with non-Gaussian noise. Then it analyzes causal discovery from event sequences. It finalizes by introducing the problem of subsampling, and different strategies to cope with this problem.