Use Randomizers to Mimic Machine Learning

Thanks for defending me on this.

1 Like

Ngl, I did kind of intend it to be based on rng, but now since people have also given their ideas, I want it to be machine learning based, as @Shdwy has talked about, as that seems like a better idea. If that makes any sense, because I’m multi-tasking on several stuff right now.

1 Like

Backpropagation.

Or if someone invents calculus in gimkit we can do gradient descent:

1 Like

That’s pretty interesting. Hopefully someone with greater coding skills than me can try and work this out.

Sorry about not noticing this, got caught up with other things. Hopefully that it works well for you!

The randomizer system is still machine learning, its not just a simulation of it. (IDK if you or a regular renamed the title because someone keeps renaming other people’s topics, but just letting whoever did know)

1 Like

Oh ok, I see now. And yeah, I was the one who changed the title

1 Like

Oh, I didnt know the title changed again. Jeffo was the one to change it this time, though, so uh… yeah. Im fine with it though.

I have a lot of experience in trying to make AI from scratch… the method, which should work I think, won’t be very effective.

1 Like

You could make an AI that learns with weights, rewards and punishments with randomness, however, that AI wouldn’t carry over with different games and AI learning is known to take a long time, even with advanced AI learning algorithms much less a simple AI learning algorithm, so unless you want the AI to restart every game, AI is not feasible in this type game development well, unless the developers add data transfers and cloud storage.

2 Likes

he didnt say that it would do that

He’s talking about something he said earlier.

what do you propose?

also interesting that jeff renamed the topic lol

1 Like

Yeah, never thought that would happen :laughing:

1 Like

Probably a neural network with backpropagation or gradient descent.

1 Like

bump

1 Like

“Better Call Saul”

1 Like