sigmoid function vs softmax function

DIFFERENCE BETWEEN SOFTMAX FUNCTION AND SIGMOID FUNCTION

二者主要的区别见于，

• softmax 用于多分类，sigmoid 则主要用于二分类；

⎧⎩⎨⎪⎪⎪⎪⎪⎪⎪⎪F(Xi)=11+exp(−Xi)=exp(Xi)exp(Xi)+1F(Xi)=exp(Xi)∑kj=0exp(Xj),i=0,1,…,k

import numpy as np
import matplotlib.pyplot as plt

def sigmoid(inputs):
    return np.exp(inputs)/(np.exp(inputs)+1)
def softmax(inputs):
    return np.exp(inputs)/sum(np.exp(inputs))

x = range(21)
sigmoid_x = sigmoid(x)
softmax_x = softmax(x)

plt.plot(x, sigmoid_x, x , softmax_x, lw=2)
plt.legend(['sigmoid', 'softmax'])

plt.show()

转载于:https://www.cnblogs.com/mtcnn/p/9421702.html