Neural Network idea?

[Gammer0909]

There's probably a video out there explaining that someone taught an AI to play minecraft, but what if we teach an AI to play bedwars?

Just a thought (I'm dumb and can't AI but hmu)

[IDoTheHax]

Alr in development

[IDoTheHax]

Nope, I lost the pc that had all of the code and didn't push the latest working code to GitHub because I'm an idiot,

When I get funded/save for a new computer I'll probably be able to do it

[IDoTheHax]

My discord is still IDoTheHax though

[Pleasntly]

Nope, I lost the pc that had all of the code and didn't push the latest working code to GitHub because I'm an idiot,

When I get funded/save for a new computer I'll probably be able to do it

Link to the latest push before it was working? would love to try it out if you could show me

[IDoTheHax]

never got it so restarting from scratch, i think thats better though

[Tewxx]

Nope, I lost the pc that had all of the code and didn't push the latest working code to GitHub because I'm an idiot,

looking foward to it

[familiar]

Imagine a bedwars bot that could incorporate machine learning training to build a neural network, and all of this training data is stored on a external server. As more and more people use the bot, they contribute to the training of the bot, and eventually, the bot becomes the best to ever have been made.

Just a thought.

[pixo2000]

If you just need videos of wins, go on youtube. Most people only post videos if they have won the game

[Gammer0909]

So basically there would be two ways of training an ai to play beadwars 1. Player Pov learning: the ai would use the data collected from other players data of them playing (this would require some sort of program running on that players computer)

Then the other is Machine Learning: where we would give points based on certain things ie: kill a player, destroy a bed, collect an item, something like so

Actually, both of those methods are Machine Learning, just different methods of getting there...

[IDoTheHax]

Yes that is true I forgot to mention that

Also accept my DC pls

[Gammer0909]

@IDoTheHax sorry forgot about this thread for a minute, what about (Deep) Reinforcement Learning? It would be an easier way to just kinda... let it run on like a pi cluster or spare PC/Laptop for like a week or two that we could then run :)

[IDoTheHax]

Sorry for the even later response lmao im currently on track to doing that, not a lot of progress since its just a test but its going somewhere

[Pleasntly]

Is this still being developed or has it been completed somewhere?