Can anyone please explain why the summation i = 1 to n is taken care while finding the derivative? For expression without regularization, the summation i = 1 to m appears in derivative equation but summation i = 1 to n is not appearing. I know derivative of w^2 is 2w but why the summation sign for i = 1 to n has disappeared?
The notation here is a little tricky.
The first cost equation (one on top) uses the w
variable as a matrix, and the summation there refers to summing up all the params inside the w
matrix. It’s the general equation of cost that applies to all the params (including w
).
On the other hand, the derivative equation (down below) is with respect to a single param w_j
inside w
. The derivative (with respect to w_j
) of all the other params in w
that is not w_j
is 0, and so they sum up to 0. The only term that is not 0 is the derivative of w_j
, and so that is the only term you are seeing.
Thank you for explanation. I understand better now.