监督学习
supervised learning, Andrew Ng doesn’t give its definition directly. He use two examples:Regression and Classification.
Still classification
More than one feature
Therefore, in my opinion, if you want your decision boundary more specific, how to choose your features must be important.
As Andrew Ng said, it turns out for some learning problems, what you really want is not to use, like three or five features. Instead, you want to use an infinite number of features. So, how do you use computer to store an infinite number of features when your computer is gonna run out of memory.->Suported vectorized machine
In the light of all of above, supervised learning is datasets plus right answer.
无监督学习
unsupervised learing-Here is the data ,can you find some instructure in the data?
one example where clustering is used is in Google News.
What Google News has done is to look for tens of thousands of news stories and then automatically cluster them together. So the news stories that all about the same topic get displayed together.
Andrew Ng also gives an example successfuly separating two mixed micro voice.