知识卡片
贝尔曼递推依受控环境核
标准MDP中Vπ(s)=Σ_aπ(a|s)Σ_s′p(s′|s,a)[r(s,a,s′)+γVπ(s′)]。这本身基于执行行动的环境动力学;只有用受混杂观测条件分布冒充该核时才产生额外因果问题,不能把所有传统Bellman方程称为关联式。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
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
12.5.5 Causal Bellman equation RL is about searching for the optimal policy, which is characterized with the Bellman equation, often written as follows: In plain words, we’re looking for a policy Π* maximizes the cumulative reward over time. Here γ is a discount rate, a value between 0 and 1, that makes sure the agent values rewards in the near future more than rewards in
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:摘录中关于贝尔曼递推依受控环境核的定义、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:12.5.5 Causal Bellman equation RL is about searching for the optimal policy, which is characterized with the Bellman equation, often written as follows: In plain words, we’re looking for a policy Π* maximizes the cumulative reward over time. Here γ is a discount rate, a value between 0 and 1, that makes sure the agent values rewards in the near future more than rewards in