Thread: My geek thread
View Single Post
Old 03-18-22, 09:29 PM
  #1808  
Flatballer
No matches
 
Flatballer's Avatar
 
Join Date: Aug 2008
Location: Eastern PA
Posts: 11,647

Bikes: two wheeled ones

Mentioned: 15 Post(s)
Tagged: 0 Thread(s)
Quoted: 1398 Post(s)
Liked 444 Times in 250 Posts
Anyone do any machine learning/reinforcement learning?

I got into it recently because there's some really impressive AI bots that can play Rocket League. I'm working first on the background to have a good foundation and then I'm gonna do some learning and practicing in open AI gym (a framework environment for RL), and then take on writing one myself. Reinforcement learning is what Alpha Go and Open AI Five (can destroy professional DOTA2 players) and other impressive game playing AI use. It's incredibly powerful and really interesting the way it works. Rocket League in particular is a real challenge for them because of the way the game is played and how open ended and strategic and stuff it is. But it's an excellent challenge to work on.
Flatballer is offline