3 ms·
From the explanations of how AlphaGo was architectured and trained: The goal being maximized isn't stone difference, rather probability of winning.
by euyyn 9y ago
From the explanations of how AlphaGo was architectured and trained: The goal being maximized isn't stone difference, rather probability of winning.