感知机学习算法

 1  # coding:utf-8
 2 import matplotlib.pyplot as plt
 3 import numpy as np
 4 
 5 def dataN(length):#生成数据
 6     x=[[1,x/100.0,x/100.0+ (x%2)*1.5 + np.random.uniform(0,1.2)] for x in range(length)]
 7     y=[(-1)**(y%2-1) for y in range(length)]
 8     x=np.mat(x)
 9     return x,y
10 
11 def percepT(x,y, iter):#感知机原始形式
12     n =np.shape(x)[1]
13     m=len(x)
14     theta=np.ones(n)
15     alpha=0.02
16     for it in range(iter):
17        l=0
18        print it  #0 1 2 3 4 5 6 7
19        for k in range(m):
20            if (x[k].dot(theta.T))*y[k]<0:
21                l=1
22                theta=np.add(theta,alpha*y[k]*x[k])
23        if l==0:
24             break
25     return theta
26 
27 def showP(x,y,theta):#作图
28     plt.figure(1)
29     color=[‘‘,‘or‘,‘ob‘]
30     for i in xrange(length):
31         plt.plot(x[i, 1], x[i, 2],color[int(y[i])])
32     theta=theta.getA()
33     plt.plot([0,length/100],[-theta[0][0],-theta[0][0]-theta[0][1]*length/100]/theta[0][2])
34     plt.show()
35     
36 length=200
37 iter=200
38 x,y=dataN(length)
39 theta=percepT(x,y,iter)
40 print theta   #[[-0.38       -0.2464      0.25287193]]
41 showP(x,y,theta)

 1  # coding:utf-8
 2 import matplotlib.pyplot as plt
 3 import numpy as np
 4 
 5 def dataN(length):#生成数据
 6     x=[[x/100.0,x/100.0+ (x%2)*1.5 + np.random.uniform(0,1.2)] for x in range(length)]
 7     y=[(-1)**(y%2-1) for y in range(length)]
 8     x=np.mat(x)
 9     return x,y
10 
11 def percepTG(x,y, iter):#感知机对偶形式
12     m=len(x)
13     theta=np.zeros(m)
14     b=0
15     alpha=0.1
16     for it in range(iter):
17        l=0
18        print it  #0 1 2
19        for k in range(m):
20            if y[k]*(np.sum([theta[j]*y[j]*x[j]*x[k].T for j in range(m)])+b)<=0:
21                l=1
22                theta[k]+=alpha
23                b+=alpha*y[k]
24        if l==0:
25             break
26     return theta,b
27 
28 def showP(x,y,theta,b):#作图
29     plt.figure(1)
30     color=[‘‘,‘or‘,‘ob‘]
31     for i in xrange(len(x)):
32         plt.plot(x[i, 0], x[i, 1],color[int(y[i])])
33     c= np.zeros(2)
34     for j in range(len(x)):
35         c=np.add(c,theta[j]*y[j]*x[j])
36     c=c.getA()
37     print c,b  #[[-0.351     0.402276]] -0.6
38     plt.plot([0,length/100],[-b,-b-c[0][0]*length/100]/c[0][1])
39     plt.show()
40 
41 length=200
42 iter=5
43 x,y=dataN(length)
44 theta,b=percepTG(x,y,iter)
45 showP(x,y,theta,b)