技術標籤：機器學習

文章目錄

• 用MLPClassifier分成三類並用矩陣演算演示了一下
• 用MLPRegressor去證實我的想法！

用MLPClassifier分成三類並用矩陣演算演示了一下

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))

用MLPRegressor去證實我的想法！

• 我證實了輸出層的啟用函式果然是恆等對映呦呦！

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

• 上面的predict的結果和我寫的result是一樣的！
• 真是太神奇啦！