Hello everyone, I have made a representation graph of implementing NN, based on Andrew’s “Forward and backward functions” drawing in week4’s “Building Blocks of Deep Neural Networks” video.
This graph uses different colors to represent how the variables flow within and between layers. I believe it should be a closer representation to the structure of week4’s assignment, including the “activation cache” and “linear cache” implementation in our code.
It would be great if someone can help me verify whether there are any mistake in my representation. Also, during the assignment I had noticed that the “b” variable seemed actually not being used in calculating “dW” and “db” in the “linear_backward” function, so I did not include “b” in the “L-cache” in my graph, can someone also confirm for me please?
In case there is no mistake in my graph, I hope this could be helpful to you guys, especially in doing the assignment and gaining intuition behind the codes. Since our assignments are mainly about filling the missing functions, not developing the whole structure from scratch, I often found myself lacking a clear structural view right after finishing the assignment, but drawing a graph like this definitely help clearing my mind a lot!
