wk4

Question 1

What does w^L_{j,k} denote

Answer 1

connection of weights from neuron k in layer L-1. To neuron j in later L

Question 2

What is z^L_j in a neural network

Answer 2

weighted input to unit j in layer L. It is the sum of the previous weights put through the previous activation function a^L-1_k + bias (b^L_j)

Question 3

what does m often mean in the context of a neural network

Answer 3

the output number of neurons

Question 4

in a fully connected multi layer perceptron. Given neuron width m and number of layers L. How many parameters are there to train

Answer 4

(m^2+m)(L-1)