All geek questions in one place

What is the meaning of three-layer NN, 300 + 100 hidden units?

I am new to NN. I was getting MNIST Handwritten Dataset. They indicated the percentage of error for each method that they adopted in tabular form. Here is the link to the page. In the NN section, they indicated how:

3-layer NN, 300 + 100 hidden units

2-layer NN, 300 latent units, standard error

What does 3 layers mean? this is

InputLayer+HiddenLayer+OutputLayer

OR

 InputLayer + 3 Hidden Layer+ outputLayer

Also what is the meaning of 300 + 100 hidden units. If you specify two numbers 300 + 100, one hidden layer should contain 300 units , and the next HiddenLayer should contain 100 units .

If so, then why 3 Layer NN?

+5
source share
1 answer

In this designation, the input layer is not taken into account, so 3 layers mean 2 hidden layers and 1 output level. So in this case:

 3-layer NN, 300+100 hidden units

means the following architecture:

 Input layer Hidden layer of 300 units Hidden layer of 100 units Output layer
+4
source

Source: https://habr.com/ru/post/1240851/

More articles:


All Articles