Candid / Profile

One day I suddenly realised a simple thing: least-squares approximation is essentially about minimising a linear combination of vectors. That is, some kind of universal approximator function can be manufactured. It's a done deal, so here is the title