One hidden layer Neural Network - Neural Network Representation

最新推荐文章于 2021-08-13 18:33:29 发布

王彩旗 edwardwangcq.com

最新推荐文章于 2021-08-13 18:33:29 发布

阅读量128

点赞数

分类专栏：人工智能 # Neural Networks and Deep Learning

本文链接：https://blog.csdn.net/edward_wang1/article/details/118487848

版权

人工智能同时被 2 个专栏收录

142 篇文章 0 订阅

订阅专栏

Neural Networks and Deep Learning

32 篇文章 0 订阅

订阅专栏

本文分享了Andrew Ng教授在Coursera课程《深度学习》中关于神经网络表示的3.2节内容，涉及输入层、隐藏层和输出层的结构，参数定义，以及2层神经网络的传统命名。理解这些概念有助于深入学习神经网络技术。

摘要由CSDN通过智能技术生成

The notes when study the Coursera class by Mr. Andrew Ng "Neural Networks & Deep Learning", section 3.2 "Neural Network Representation". Share it with you and hope it helps!

------------------

Figure-1 shows names of different parts of Neural Network.

It has Input layer, hidden layer and output layer
The term hidden layer refers to the fact that in a training set the true values for these nodes in the middle are not observed.
An alternative notation for the input x is $a^{[0]}$ where a stands for activation
The hidden layer generates some set of activations called $a^{[1]}$ . The first unit generates value $a^{[1]}_{1}$ , the second unit generates value $a^{[1]}_{2}$ and so on. So here $a^{[1]}=\begin{bmatrix} a^{[1]}_{1}\\ a^{[1]}_{2}\\ a^{[1]}_{3}\\ a^{[1]}_{4} \end{bmatrix}$
The output layer generates value $a^{[2]}$ which is just a real number. And $\hat{y}=a^{[2]}$
One funny thing is this neural network here is called a 2-layer NN in convention. The reason is we don't count input layer.
The hidden layer will have parameters associated with it: $W^{[1]} (4\times3)$ and $b^{[1]} (4\times1)$
The output layer will have parameters associated with it: $W^{[2]}(1\times4)$ and $b^{[2]}(1\times1)$

<end>