机器学习-线性分类器

#线性感知机算法

import numpy as np

def Perceptron_1(X_train,y_train,eta = 1):
    d = np.ones(len(X_train))
    X_train = np.c_[X_train,d.T]
    w = np.zeros(X_train.shape[1])
    
    converged = False
    it = 0
    while not converged:
        converged = True
        for i in range(len(X_train)):
            x = X_train[i]
            y = y_train[i]
            if y*(x.dot(w))<=0:
                w = w+eta*y*x
                converged = False
        print('Iter',it,w)
        it = it+1
    return w
x = [[0,0],[2,2],[2,0],[3,0]]
y = [-1,-1,1,1]

x = np.array(x)
y = np.array(y)

w = Perceptron_1(x,y)
print("trained",w)

输出为：

Iter 0 [ 2.  0.  0.]
Iter 1 [ 2. -2. -1.]
Iter 2 [ 2. -2. -1.]
trained [ 2. -2. -1.]

注：改变坐标点的顺序会使结果改变，是因为问题有多种解。
例如：

x = [[0,0],[2,0],[3,0],[2,2]]
y = [-1,1,1,-1]

x = np.array(x)
y = np.array(y)

w = Perceptron_1(x,y)

输出为：

Iter 0 [ 0. -2. -1.]
Iter 1 [ 0. -4. -1.]
Iter 2 [ 2. -4.  0.]
Iter 3 [ 2. -4. -1.]
Iter 4 [ 2. -4. -1.]