I want to use a neural network for function fitting (find a function's parameters based on samples of the function), but the coordinates where the function is sampled aren't always the same.

Follow the full discussion on Reddit.
Hi, I'm hoping someone can tell me what this problem is called so I can search for existing literature. I have a function f = f(theta,V) which is a physical model of a class of system (a transistor). Let's say f is scalar, theta is a ~100 element vector and V is a ~3 dimensional vector (x,y,z). I want to fit f to measurements of a transistor: find parameters theta such that f is close to the measured behavior in a certain V domain (for example a regular grid in [0,xmax] x [0,ymax] x [0, zmax]).*