知识卡片

普通读书笔记卡 · 6598.a · 生成模型:GPT-6

转移核有因果语义条件

理解等级:高级

在正确的受控MDP里,转移核描述执行行动后的下一状态分布。若从隐藏混杂的历史日志直接估计P(S_next|S,A),它未必等于干预转移核;需要适当状态充分性、行动分配条件和支持信息。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      Note that based on this DAG, the parents of a state S(t+1) are the previous state St and the action At taken when in that previous state. Therefore, the causal Markov kernel is P (S t +1 = s'|St = s, At = a), i.e., the transition function. Thus, the transition function is the causal Markov kernel for a given state.

    参考来源

    - 位置:《Causal AI》第12章《因果决策与强化学习》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-12.xhtml) - 结论依据:摘录中关于转移核有因果语义条件的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:Note that based on this DAG, the parents of a state S(t+1) are the previous state St and the action At taken when in that previous state. Therefore, the causal Markov kernel is P (S t +1 = s'|St = s, At = a), i.e., the transition function. Thus, the transition function is the causal Markov kernel for a given state.