用MLPClassifier分成三类并用矩阵演算演示了一下+用MLPRegressor去证实我的想法_mlpclassifier sequential mlpregressor 关系

用MLPClassifier分成三类并用矩阵演算演示了一下

from sklearn.neural_network import MLPClassifier
import numpy as np
X = [[0., 0.], [1., 1.],[3,3]]
y = [0, 1,2]
clf = MLPClassifier(solver='lbfgs', alpha=1e-5,
                    hidden_layer_sizes=(2,),
                    random_state=1, activation='identity')
clf.fit(X, y)

print (clf.predict([[1.2,1]]))



a = np.matrix(clf.coefs_[0])
b = np.matrix(clf.coefs_[1])
theta1 = np.matrix(clf.intercepts_[0])
theta1 = theta1.T
theta2 = np.matrix(clf.intercepts_[1])
theta2 = theta2.T
x=np.matrix([[1.2],[1]])
result = np.exp(b.T * (a.T*x+ theta1)+theta2)
print (result/sum(result))
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

用MLPRegressor去证实我的想法！

• 我证实了输出层的激活函数果然是恒等映射呦呦！

from sklearn.neural_network import MLPRegressor
import numpy as np
X = [[0., 0.], [1., 1.],[3,3]]
y = [0., 1.,2.]
clf = MLPRegressor(solver='lbfgs', alpha=1e-5,tol = 1e-6,
                    hidden_layer_sizes=(2,),
                    random_state=1, activation='identity')
clf.fit(X, y)

print (clf.predict([[3,3]]))



a = np.matrix(clf.coefs_[0])
b = np.matrix(clf.coefs_[1])
theta1 = np.matrix(clf.intercepts_[0])
theta1 = theta1.T
theta2 = np.matrix(clf.intercepts_[1])
theta2 = theta2.T
x=np.matrix([[3.],[3.]])
result = b.T * (a.T*x+ theta1)+theta2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21

• 上面的predict的结果和我写的result是一样的！
• 真是太神奇啦！