Homework_Week8_Coursera【Machine Learning】AndrewNg、Unsupervised Learning
- 1 For which of the following tasks might K-means clustering be a suitable algorithm? Select all that apply.
- 2 Suppose we have three cluster centroids u=...
- 3 K-means is an iterative algorithm, and two of the following steps are repeatedly carried out in its inner-loop. Which two?
- 4 Suppose you have an unlabeled dataset....
- 5 Which of the following statements are true? Select all that apply.
1 For which of the following tasks might K-means clustering be a suitable algorithm? Select all that apply.
解析
A 从不同的新闻网站发现主要的主题、明显是一个聚类算法
B 从网站的使用模式,发现用户的不同群体、
C 有很多EMAIL,你想要判断是否是垃圾邮件、错误,明显是监督学习,且非聚类
D根据历史记录,猜测明天是有雨还是晴天、错误,属于回归问题
答案
AB
2 Suppose we have three cluster centroids u=…
解析
假设有三个族的中心u1,u2,u3。我们有一个训练样本x(i),感觉和u3有重合、
算距离根号下a2+b2 ,1根号5 2是9 3是根号2
距离最短
答案
D
3 K-means is an iterative algorithm, and two of the following steps are repeatedly carried out in its inner-loop. Which two?
解析
KMEAN是一个可迭代算法,并且以下两步会直接导致内部循环,哪两个呢?
答案
AC
4 Suppose you have an unlabeled dataset…
解析
假设你有未标签数据集、你跑了KMEANS用50钟不同的类别、有哪些推荐的方法给者50种类别使用
这节课学的知识点
答案
C
5 Which of the following statements are true? Select all that apply.
解析
A 选正确的、担心Kmean会困在糟糕的局部最优解,一个优化方法就是随机生成初始化,没错
B 自从KMEAN是一个无监督学习算法,它不会过拟合数据、所以永远都是集合越多越好、肯定不是,这里参考课上讲的肘部拐点选择法、横轴分类数量 纵轴损失函数j
C正确、确实一些数据集中的正负样例摸棱两可的
D 错误怎么会都是从0向量开始呢,应该是随机
答案
AC