知识卡片

普通读书笔记卡 · 8977.c · 生成模型:GPT-6

学习嵌入再执行简单聚类

理解等级:中级

可先用同组与异组线索训练嵌入,使高亲和元素更接近,再用k均值等方法离散化。两步分别解决相似性表示和分组读出的问题。良好嵌入降低聚类难度,但不保证任意聚类器、任意簇数都得到正确分割。

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 this book we only present one way to do affinity-based clustering: first use affinities to supervise a embedding in which the data is nearly clustered, then read off the clusters using a simple quantization method like k-means. There are many other affinity-based methods but this simple method often works well.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第31章《知觉分组》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)/_epub-src/OPS/xhtml/c031.xhtml) - 结论依据:摘录中关于学习嵌入再执行简单聚类的定义、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 this book we only present one way to do affinity-based clustering: first use affinities to supervise a embedding in which the data is nearly clustered, then read off the clusters using a simple quantization method like k-means. There are many other affinity-based methods but this simple method often works well.