TensorFlow常用激活函数

最新推荐文章于 2022-07-23 00:13:17 发布

原创最新推荐文章于 2022-07-23 00:13:17 发布 · 345 阅读

1 ·

CC 4.0 BY-SA版权

--------Tensorflow 学习-------- 专栏收录该内容

5 篇文章

订阅专栏

本文探讨了神经网络中关键的激活函数，包括阈值函数、Sigmoid、Tanh、RELU和Softmax，通过代码实现与可视化，展示了这些函数在神经元活动与输出之间的转换特性。

threshold

import os
os.environ['TF_CPP_MIN_LOG_LEVEL']='2'
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
 
def threshold(x):
    cond = tf.less(x, tf.zeros(x.shape, dtype=x.dtype))
    res = tf.where(cond, tf.zeros(x.shape), tf.ones(x.shape))
    return res

h = np.linspace(-1,1,50)
out = threshold(h)

with tf.Session() as sess:
    y = sess.run(out)
    plt.xlabel('Activity of Neuron')
    plt.ylabel('Output of Neuron')
    plt.title('Threshold Activation Function')
    plt.plot(h, y)
    plt.show()

在这里插入图片描述

Sigmoid

import os
os.environ['TF_CPP_MIN_LOG_LEVEL']='2'
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
 
h = np.linspace(-10,10,50)
out = tf.sigmoid(h)

with tf.Session() as sess:
    y = sess.run(out)
    plt.xlabel('Activity of Neuron')
    plt.ylabel('Output of Neuron')
    plt.title('Sigmoid Activation Function')
    plt.plot(h, y)
    plt.show()

在这里插入图片描述

Tanh

import os
os.environ['TF_CPP_MIN_LOG_LEVEL']='2'
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
 
h = np.linspace(-10,10,50)
out = tf.tanh(h)

with tf.Session() as sess:
    y = sess.run(out)
    plt.xlabel('Activity of Neuron')
    plt.ylabel('Output of Neuron')
    plt.title('Tanh Activation Function')
    plt.plot(h, y)
    plt.show()

在这里插入图片描述

RELU

import os
os.environ['TF_CPP_MIN_LOG_LEVEL']='2'
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
 
h = np.linspace(-10,10,50)
out = tf.nn.relu(h)

with tf.Session() as sess:
    y = sess.run(out)
    plt.xlabel('Activity of Neuron')
    plt.ylabel('Output of Neuron')
    plt.title('RELU Activation Function')
    plt.plot(h, y)
    plt.show()

在这里插入图片描述

Softmax

import os
os.environ['TF_CPP_MIN_LOG_LEVEL']='2'
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
 
h = np.linspace(-5,5,50)
out = tf.nn.softmax(h)

with tf.Session() as sess:
    y = sess.run(out)
    plt.xlabel('Activity of Neuron')
    plt.ylabel('Output of Neuron')
    plt.title('Softmax Activation Function')
    plt.plot(h, y)
    plt.show()

在这里插入图片描述