Dot product to calculate z -- Details

When creating a post, please add:

Within Neural Network Layer course, there is explaination that how each node in size of layer 1 is calculated with using the logistics regression (using the sigmoid function). When we are calculating below,

Let’s say

w1 = 0.001

And since our input(x) was

[197, 184, 136, 214]

z will be

(197 * 0.001) + (184 * 0.001) + (136 * 0.001) + (214 * 0.001)  + 3 
= 7.31 - b
= z

and we are using the sigmoid function to get the result for the first nuron, which will be like below

1/ (1 + e^(-z))

and that g(z) was 0.3 for the example right?

Did I understand correctly?

The four numbers you have for x, are those four separate examples where there is only one feature?

Or is that one example that has four features?

One example with four features Or let’s say there were 2 hidden layer l1 and l2. If I say that the l1’s had 4 nodes, so it returned 4 numbers from the each node. And the current layer’s node size is 3, then would my calculation will be correct?

Every layer must have the same number of weights as there are features. This is because the weights and features are multiplied together.

1 Like