Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
shghs
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
4 ms
·
1.
▲
by
shghs
9y ago
The main reason AlphaGo Zero learns so much faster than its predecessors is because it uses temporal-difference learning.[1] This effectively removes a huge amount of the value network's state space for the learning algorithm to sear