← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6677.b · 生成模型:GPT-6

主成分符号和重根方向可不唯一

理解等级:中级

同一主轴及其相反方向表达相同子空间,图像左右或上下翻转不代表模型质量变化。若若干特征值相等,相应子空间内基方向还可能旋转;比较两次结果应关注子空间及下游任务,而非只要求每个坐标符号一致。

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

      When you reduce the dimensionality of 3D points (vectors) to 2D, it’s like taking a picture of that 3D cloud of points. The result may look like a picture on the right or the left of figure 4.4, but it will never tip or twist to a new angle. The x-axis (axis 0) will always be aligned along the longest axis of the point cloud

    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:摘录中关于主成分符号和重根方向可不唯一的定义、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:When you reduce the dimensionality of 3D points (vectors) to 2D, it’s like taking a picture of that 3D cloud of points. The result may look like a picture on the right or the left of figure 4.4, but it will never tip or twist to a new angle. The x-axis (axis 0) will always be aligned along the longest axis of the point cloud