Sentence clear concept: no programming to develop the machine what to do, the machine has the ability to learn.
The three most basic of machine learning algorithms: decision trees, linear regression, K-means clustering.
Supervision and unsupervised learning
Zhou Zhihua following explanation by knocking watermelon Case:
Supervised learning is to listen to the sound of knocking watermelon judge the quality of the process have knowledgeable people tell you this melon is good or bad, to different voices labeled good or bad melon label, slowly learning the relationship between sound features, and finally through feature can be predicted by the model.
Supervision model can be divided into classification and regression model. Classification model to predict the label is a categorical variable, the regression model predicts tags for numeric variables.
Unsupervised learning is not knowledgeable people tell you that the melon is good or bad, can only make the sound characteristics classification (cloudy, crisp, dull), tag the next time get a new melon listen to the voice belongs to which.
For example, K-means clustering
