← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6599.c · 生成模型:GPT-6

博弈反事实遗憾按信息集定义

理解等级:高级

博弈CFR在信息集上比较替代行动的反事实价值,并由对手与机会的到达概率加权,迭代最小化局部遗憾。它不是事后知道对手底牌后重选最佳动作;其均衡保证还依赖相应博弈类别与自博弈条件。

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

      12.6.2 Counterfactual regret minimization In chapters 8 and 9, I introduced regret as a counterfactual concept. We can further clarify the idea now that we have introduced the language of decision-making; regret is the difference between the utility/reward that was realized given a specific action or set of actions, and the utility/reward that would have been realized had another action or set of actions been

    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:摘录中关于博弈反事实遗憾按信息集定义的定义、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:12.6.2 Counterfactual regret minimization In chapters 8 and 9, I introduced regret as a counterfactual concept. We can further clarify the idea now that we have introduced the language of decision-making; regret is the difference between the utility/reward that was realized given a specific action or set of actions, and the utility/reward that would have been realized had another action or set of actions been