5 ms·
It seems that AlphaGo has got even better since when it defeated Lee Sedol. This time, AlphaGo is undefeated against other top players. However, it also gives o
by nullnilvoid 10y ago
It seems that AlphaGo has got even better since when it defeated Lee Sedol. This time, AlphaGo is undefeated against other top players. However, it also gives other players a chance to practice against AlphaGo and finds weaknesses in AlphaGo, which is super important for any competition.
- 21 10y agoIt would be super-weird if AlphaGo became worse that the previous version.
- daveguy 10y agoNeural networks have a bad habit of regressing. When training continues they tend to replace earlier skill with newer skill.
- 21 10y agoThe team would surely detect that. In the original paper they even mention this issue, and as a safe-guard they also test later networks against previous versions.
- daveguy 10y agoTrue, they are aware and monitor for it. Just pointing out that it's not a super weird thing to happen with a NN. In fact it's a super normal thing that has to be continuously monitored.