[ 机器学习 – 吴恩达 ] 介绍 | 1-4 非监督学习([machine learning – Wu Enda] introduction | 1-4 unsupervised learning)

聚类算法

例子:谷歌新闻,基因分类,组织计算机集群,社交网络分析,市场划分,天文数据分析

鸡尾酒会问题

[W,s,v] = svd((repmat(sum(x.*x,1),size(x,1),1).*x)*x’)

问题1:以下示例中,你会用非监督学习算法解决哪一个?
答案: BC
A: 给定标记为垃圾或非垃圾的邮件,学习一个垃圾邮件分类器
B: 给定在网上找到的一组文章,将它们分为关于同一故事的一组文章
C:给定一个客户数据库,自动发现细分市场,并将客户分为不同的市场划分。
D:给定一个诊断为患有或未患糖尿病的患者数据集,学习将新患者分为是否患有糖尿病。

————————

clustering algorithm

Examples: Google News, gene classification, organization, computer clustering, social network analysis, market segmentation, astronomical data analysis

Cocktail Party Problem

[W,s,v] = svd((repmat(sum(x.*x,1),size(x,1),1).*x)*x’)

< strong > question 1: < / strong > which of the following examples would you use unsupervised learning algorithm to solve?
< strong > answer: < / strong > BC
A: Given a message marked as spam or non spam, learn a spam classifier
B: Given a group of articles found on the Internet, divide them into a group of articles about the same story
C: Given a customer database, we can automatically find market segments and divide customers into different market segments.
D: given a dataset diagnosed with or without diabetes, we divided the new patient into diabetes.