知识卡片
重要性权重纠正采样分布
重要性采样从便于采样的提议分布取样,再用目标密度与提议密度之比修正。提议分布必须覆盖目标的相关支持,权重退化会降低有效信息量。不能把“给样本乘目标概率”当作适用于所有提议分布的通用公式。
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
- · 概率生成建模入门
原书出处 · 1 处
- · 查看原书依据
These methods generate outcomes from a joint probability distribution and then weight them according to their probability in the target distribution. We can then use the weights to do weighted averaging via Monte Carlo estimation. Popular variants of this kind of inference include importance sampling and inverse probability reweighting, the latter of which is popular in causal inference and is covered in chapter 11.
参考来源
- 位置:《Causal AI》第02章《概率生成建模入门》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-2.xhtml)
- 结论依据:摘录中关于重要性权重纠正采样分布的定义、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:These methods generate outcomes from a joint probability distribution and then weight them according to their probability in the target distribution. We can then use the weights to do weighted averaging via Monte Carlo estimation. Popular variants of this kind of inference include importance sampling and inverse probability reweighting, the latter of which is popular in causal inference and is covered in chapter 11.