I have similar architecture like in image:CNN.
I don't understand how to calculate gradient of filter F.
I found these equations(source):
Gradient and delta,
where first equation calculate gradient of weight Fab from filter F and second equation calculate deltas of input layer and
E - total error,
N - input width,
M - input height,
l - layer. 0 - first, l = L last
x - input
y - output
o - activation function
o' - derivate of activation function,
but I don't understand how to calculate ∂E/∂yijl.
How can I can calculate gradient of F?