Quote:
Originally Posted by goofyballer
But why? As a programmer, I don't take "we taught it to learn" as an explanation. Without some a priori knowledge of its environment (or desire or objective) baked in, there's no reason for it to do anything other than sit there.
It's taught the rules of the game, which includes the objective, but is given no "expert knowledge" to start with.
For example, a typical chess program would be programmed with knowledge of how to evaluate positions (usually aided by a chess master), opening books etc. By contrast, with AlphaZero everything is "learned" from scratch by playing massive numbers of games against itself to see what works and what doesn't.
I'm sure I'm oversimplifying it but that's basically the gist of it from what I understand (and I haven't watched the doc yet).
Last edited by TimTimSalabim; 01-30-2018 at 02:29 PM.