There seems to be a sort of “occam’s razor”-alike in machine learning: systems with the fewest human-coded assumptions are winning.
Conversation
Replying to
I'd be careful with the broad assertion - I've seen both ways depending on the domain. Esp when geometry is involved.
1