I was going through CNN's and found that padding argument should be set to "Valid" if i need no padding and "Same" if i need padding. But, it doesn't make any sense to me. Why can't keras development team just put "Padding = True", is there a specific reason to choose this convention?
1 Answer
$\begingroup$
$\endgroup$
Using valid will essentially use as much of your input as possible, such that the dimensions continue to work. This means there is a chance some input will be trimmed (removed).
same on the other hand, will add padding to allow e.g. the number of filters/convolutions you specify to be applied.
The reason not to simply have padding=True as an option, is that there are more than two options (see snippet from the docs below). Also, that would not be very explicit: using same or valid makes it crystal clear, what the result should be using the given approach.
This is what the documentation says:
- padding: One of "valid", "causal" or "same" (case-insensitive). "valid" means "no padding". "same" results in padding the input such that the output has the same length as the original input
Take a look at this nice answer with a simple example.
