SVM machine learning algorithm

1, the knowledge points

"" " 
SVM, also known as support vector machines: the optimization core to strike a point to maximize the distance from the plane, mainly to solve binary classification problem 
    y = wx + b, and yi * y (i)> 0 constant Founded 
    thought: min {the argmax [Yi * (WX + B)] / || W ||} 
        min [Yi * (WX + B)]: a straight line is obtained from the sample point nearest 
        argmax: obtaining a sample point nearest the maximum distance value 

    optimization goal: max (w, b) { 1 / || w ||}, constraints yi * y (i)> = 1, the minimum Lagrangian strike 
    all points (this point is also on the boundary called support vector), alpha is not equal, not points on the boundary, alpha equals 0 
    SVM is a real role to play with you, alpha is not the point zero 

    nuclear transformation: solving linear inseparable problem, kernel functions are: radial Basis function , Gaussian function, rbf function 
"" "

2, the code

 

3, the objective function

4, optimization goals