Top Page | Upper Page | Contents | About This Site | JAPANESE

Supervised and Non-Supervised Learning

In Machine Learning , "Supervised" and "Non-Supervised" are often used.

Many methods are used as one side. But Neural Network and MT method are used for both.

Supervised Learning

If there is the data "Y" and "X". It is Supervised Learning.

Regression Analysis , Discriminant Analysis , Decision Tree and so on.

Example

• There are two variables, X1, X2, color.
  
• There could be a rule
  "if X1 > 4 then orange", "if X1 <= 4 then blue".
  This is supervised learning.
  
• This output is used for prediction. For example,
  "It will be orange because X1 = 5 and X2 = 6"

Non-Supervised Learning

Non-Supervised Learning is used to find the characteristics of the data.

Cluster Analysis , Principal Component Analysis , Self Organizing Map and so on.

Example

• There are X1 and X2.
  
• By grpah, there is a rule that
  "X1 = 4 is border line of two groups"
  This is non-supervised leraning.
  
• This output is used for prediction. For example,
  "It will be the group of right side because X1 = 5 and X2 = 6"

NEXT Reinforcement Learning