Scholarlink

In Chap. 10 , we showed how to overcome the curse of dimensionality for counters by imposing independence of the counter and label variables corresponding to different subsystems. This chapter lays the foundation for incorporating more flexibility with respect to the independence assumptions imposed between the counter and label variables. Specifically, we will use the global Markov property to define the conditional independence relationships between these variables, specified by a modeler via information structure graphs. The choice of information structure affects the resulting approximation as well as the computational requirements. The information structure perspective generalizes the subsystem setting of Chap. 10 .

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

Information Structure

  • 章节

  • 出版模式：

    开放获取

  • 发表日期：

    2025年

查看全文
  • Antonius B. Dieker,
  • Steven T. Hackman

摘要

In Chap. 10 , we showed how to overcome the curse of dimensionality for counters by imposing independence of the counter and label variables corresponding to different subsystems. This chapter lays the foundation for incorporating more flexibility with respect to the independence assumptions imposed between the counter and label variables. Specifically, we will use the global Markov property to define the conditional independence relationships between these variables, specified by a modeler via information structure graphs. The choice of information structure affects the resulting approximation as well as the computational requirements. The information structure perspective generalizes the subsystem setting of Chap. 10 .