Machine learning Preface
Definition
- T: Task
- E: Experience
- P: Performance
- Sequence: T -> E -> P
Supervised learning
Definition
- Give the right answer to each example of the data set(called training data).
Type
- Regression: get the continuous values
- Classification: get the discrete values like 0, 1, 2, 3 and so on
application scenarios
-
Regression: predict the price of the house based on the square, location of the house
- house price
-
Classification:
- Tumor prediction
- Spam filter
Unsupervised learning
Type
- Cluster algorithm
application scenarios
- Google news: get lots of related news in the Internet and put them in one set of URL.
- Social network: find the common friends.
- Market segmentation: We all know the data, but we don't know the what kinds of market segmentation, so let unsupervised learning to deal with it.
- Extract human voice from records: you know, there are some noise in these records, we need to get the human voice, so we let cluster algorithm to deal with.