机器学习算法导论---kernel_svm

import numpy as np
from machine_learning.lib.svm_smo import SVM

class KernelSVM(SVM):
    def __init__(self,kernel=None):
        self.kernel=kernel

    def get_K(self,X_1,X_2):
        if self.kernel==None:
            return X_1.dot(X_2.T)
        m1,m2=len(X_1),len(X_2)
        K=np.zeros((m1,m2))
        for i in range(m1):
            for j in range(m2):
                K[i][j]=self.kernel(X_1[i],X_2[j])
        return K

    def fit(self,X,y,N=10):
        K=self.get_K(X,X)
        self.smo(X,y,K,N)
        self.X_train=X
        self.y_train=y

    def predict(self,X):
        K=self.get_K(X,self.X_train)
        return np.sign(K.dot(self.Lambda*self.y_train)+self.b)