知识卡片

普通读书笔记卡 · 6679.c · 生成模型:GPT-6

随机投影和非负分解目标不同

理解等级:中级

随机投影用随机映射在一定条件与误差范围内近似保留点间距离;NMF将非负数据分成非负因子,可能更便于解释组成。二者不是与SVD完全等价的实现,是否使用取决于误差、约束、速度与下游目标。

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

      SVD is, by far, the most popular way to reduce the dimensions of a dataset, making LSA your first choice when thinking about topic modeling. However, there are several other dimensionality reduction techniques you can also use to achieve the same goal. Not all of them are even used in NLP, but it’s good to be aware of them. We’ll mention two methods here: random

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第04章《从词数到语义分析》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-4.html) - 结论依据:摘录中关于随机投影和非负分解目标不同的定义、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:SVD is, by far, the most popular way to reduce the dimensions of a dataset, making LSA your first choice when thinking about topic modeling. However, there are several other dimensionality reduction techniques you can also use to achieve the same goal. Not all of them are even used in NLP, but it’s good to be aware of them. We’ll mention two methods here: random