Convolutional operation: weighted average

86 Views Asked by At

I'm reading the book "Deep Learning by Ian Fellow".

At chapter 9, page 327, there is the following formula:

$$s(t)=\int x(a)\omega (t-a)da$$

This operation is called convolution, and they explain it that we apply a weight, $\omega(a)$, to a value $x(a)$ at every moment.

I don't understand why they subtract $a$ from $t$ in $\omega(t-a)$.

Why they subtract $a$ from $t$?