HackerTube 中文?

Robocalypse Now: Using Deep Learning to Combat Cheating in Counter-Strike: Global Offensive

GDC Festival of GamingYouTube ↗

Further reading 1

Where does learning get a signal without an answer key? →Start with an unlabeled patch of an image, then switch to a game that requires a sequence of actions. Neither has an answer key, but finding structure and choosing a move need different signals.This video helps answer · Which two kinds of replay share a not-guilty label?

Caption notes

Counter-Strike’s Overwatch jurors watch eight anonymized rounds and convict only when confident. Their verdicts supply learning labels with a blind spot. Subtle cheating can receive the same verdict as skillful play.

Community transcript ↗
Original passages · 2
Only convict them if you believe there is evidence beyond a reasonable doubt.
Robocalypse Now: Using Deep Learning to Combat Cheating in Counter-Strike: Global Offensive · ▶ 23:34–23:38
Overwatch doesn't catch subtle cheaters. Because we ask players, like, don't convict if you're not sure
Robocalypse Now: Using Deep Learning to Combat Cheating in Counter-Strike: Global Offensive · ▶ 52:16–52:21

Hacker News comments

View original on HN ↗