DATA100-L24: Clustering
Contents
introduction to clustering
no label at all 😢
K-means clustering
K-Means vs KNN
minimizing inertia
convex?? 损失函数不一定凸，梯度下降难顶 how to see which one is better ❓
但是找到全局最优解非常困难
agglomerative clustering
演示见上面链接以及lec code！
和CS61B的minimum spanning tree类似，每次合并两个最近的点，直到终止条件
outlier 有时忽略处理或者自成一类
picking K
Smax？
can s be negative?
summary
