Machine learning - what are machine learning, supervised learning, and unsupervised learning

What is machine learning

Let the computer be programmed without explicit , Have the ability of self-learning , For something , Computers learn from experience , And the better .

Supervised learning and unsupervised learning

1. Supervised learning
Today machine learning creates 99% The economic value of is through a kind of machine learning , That is supervised learning ： Through a large number of known input and output matching data , Let the computer learn the rules , Thus, a reasonable output prediction can be made for a new input .
such as , We have a lot of different characteristics （ area 、 Location 、 toward 、 Developers, etc ) Price data of the house , By learning these data , Predict the price of a house with known characteristics , This is called Return to learning （Regression learning）, That is, the output result is a specific value , Its prediction model is a continuous function . Another example is that we have a large number of emails , Every email has been marked as spam . By learning these marked email data , Finally, we get a model , This model is for new mail , It can accurately determine whether the email is spam , This is called Classified learning (Classfication learning), That is, the output result is discrete , That is, either output 1 It means spam , Or output 0 Means it's not spam .
2. Unsupervised learning
After supervised learning , The most widely used form of machine learning is unsupervised learning ： Through a large amount of unmarked data , To analyze the inherent characteristics and structure of the data itself .
such as , We have a lot of historical information about users' shopping , Analyze different categories of users from the data . In response to this question , We can finally divide into several categories ？ What are the characteristics of each category ？ We didn't know in advance . This is called clustering (Clustering) . Special attention should be paid to the difference between classification and supervised learning , The classification problem is that we already know which categories ; Clustering problem , We don't know which categories there are before analyzing the data . And classification problem is to choose one of the known answers , The answer to the clustering problem is unknown , We need to use algorithms to mine the characteristics and structure of data .