Machine Learning Quick Reference

Deep neural network notation

The explanation of the notation is as follows:

• l: Number of layers is 4
• n^[l]: Number of nodes in layer l 

For the following architecture, this is as follows:

• n ^[0]: Number of nodes in input layer, that is, 3
• n ^[1]: 5
• n ^[2]: 5
•  n ^[3]: 3
• n ^[4]: 1
• a ^[l]: Activations in layer l:

As we already know, the following equation goes through the layers:

z = w^TX + b

Hence, we get the following results:

•  Activation: a = σ(z) 
• w^[l]: Weight in layer l
• b^[l]: Bias in layer l