Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: THEend8_
Instructions: Answer the following questions and submit your answers on Blackboard in a single pdf file
named written.pdf. It is fine to photograph hand-written answers or type answers in MS Word, as long
as you collate your answers and convert them into this single pdf file before submission. It is also fine
to use Python to perform intermediate computations, as long as you show code snippets to explain your
reasoning. Questions begin on the following page.
1. Consider a neural network y = f(w, x) with the following form:
v = conv(w, x)
y = tanh X
m,n
vm,n!
where w, x, and v are 2D arrays, and y is a scalar. The v = conv(w, x) operation is defined as follows:
vm,n =
X
i,j
wi,jxm+i,n+j
The indices i, j range over the rows and columns of w. The result vm,n is only defined for indices
m, n where xm+i,n+j will not produce an array out-of-bounds error. In other words, the shape of v
will be smaller than the shape of x.