Help to find and build right neural network

Hi! I have a some data which i need to classify. Basically let's imagine an array of [alot,100], where alot - samples, and 1:100 are Y data for some function (X is linspace). I need to find in this array samples that have similar shape of this function. This function can be of any shape at any sample. When I say "similar shape" I mean that shapes of the functions could need to be rescaled and shifted to be found similar. Or not. So, that's why i can't use correlation analysis. What type of neural network can be smart enought to handle this task ? It is almoust like image recognition, but each sample have to be compared with all others. So, I have this array. I want to build some smart neural network, which will be trained by pairs of samples that have similar shape (found at manual mode). I hopefully to provide a result X with X[alot,alot] = 0..1 is a measure of similarity between pair of samples Some ideas?

Best regards,