I'm Johan Eliasson aka "Nebol" aka "Gigante". A Swedish citizen and a software developer. Usually busy with work/hobby stuff and not really a youtuber at all, but now and then a video MIGHT pop up...
Please don’t waste my time by uploading this. I would really have to strain to find a video less interesting than this. Granted, i’ve only watched 10 seconds of it.
The neural network has one big unresolved issue. We want randomness to bring order, but each time we add up to 10% chaos to the gene to modify it. This means that we do not improve the situation 100%, but hope for chance and at the same time we spend time on simulations. At the same time, I have no use for computer logic, which is familiar to us, because only mathematics reigns here. This is the problem that we have gone beyond elementary logic, we have abandoned optimization, cutting off unnecessary actions, speeding up the result, and so on. We are happy with the way of the 1970s, and we have to make our own intelligence in 2022 times times better than the 70s
I am very much fascinated by reinforcement learning and it's applications. Though I am not a CS student but I do want to work in this field. I have done some Deep Q learning projects using gym environment. I am wondering how can I make these kind of environments at my own. What tools, language do I need. I am just asking for creating the environment like cars, racetracks sensors etc. Can someone help me with it?
Can someone please make a sandbox game with procedural self learning npcs like this? If the game automatically singled out the behavior-patterns that scores the highest by it self like an "AI" (to keep it simple) it would be interesting what sort of gameplay that could be integrated with it. Imagine dwarf-fortress and the dwarves were learning.
Holy shit, the track from that game... i played that game on my shitty laptop like 7 years ago :(( I've wanted to find that game for years so i give you my thankyou for finding it
CS16 had a really interesting single player bots that could improve the more they played (supposedly). Now imagine other games where AI can actually improve just as you do... Insane opportunities. No more stupid "easy - normal -hard -death march" with fake difficulty by giving stupid ai cheats.