知识卡片

普通读书笔记卡 · 6654.e · 生成模型:GPT-6

降维表示在压缩与保真之间取舍

理解等级:中级

把高维统计向量映射到较少维可提取某些主题结构并降低处理成本,但也会丢信息。二维图主要用于观察,图上靠近不一定保留所有原空间关系;不能把任意投影都当能无损保留语义的嵌入。

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

      In chapter 3, you will learn how to consolidate words into a smaller number of vector dimensions to deal with the curse of dimensionality. You may even be able to turn the curse into a blessing by using all those dimensions to identify the subtle things you want your NLU pipeline to understand. You project vectors onto each other to determine the distance between each

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第01章《会读写的机器与自然语言处理》(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-1.html) - 结论依据:摘录中关于降维表示在压缩与保真之间取舍的定义、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:In chapter 3, you will learn how to consolidate words into a smaller number of vector dimensions to deal with the curse of dimensionality. You may even be able to turn the curse into a blessing by using all those dimensions to identify the subtle things you want your NLU pipeline to understand. You project vectors onto each other to determine the distance between each