Replays are in replays.zip in supplementary data
Is there a link to that? A video of this would be interesting to watch.
Grandmaster level in StarCraft II using multi-agent reinforcement learning
101–110 of 321 posts
Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning
#102Earlier quoted context omitted.
> The speed at which AlphaStar overtook previous AIs seems to me to be nearly unprecedented in AI research. Is it not simply the case that, before AlphaStar, very little money and effort was being put into developing AIs for Starcraft 2?
That depends on what you consider 'very little'. SC has had annual computer tournaments and research was published on it routinely. Few computer games see nearly as much money and effort put into developing AIs for them, yet the curve of progress was not exactly impressive.
Well, I don’t have any idea of absolute amounts of money, but I’d be more interested in relative amounts of money before and after DeepMind.
Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning
#103Earlier quoted context omitted.
Or connect the computer to human neurons directly.
You'd be able to save the time it takes a nerve impulse from head to hand that way. It's not much but apparently it can be significant enough that in some DotA games shorter players are disproportionately successful in the roles requiring the fastest reflexes.
Is that supposed to be a joke? DotA doesn't rely on reflexes nearly enough for that to matter.
Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning
#104There has always been the issue of interface when playing videogames AI vs human. Either give the human a brain-computer interface or give the AI a mouse, keyboard, monitor, robot hands and a camera. Anything else seems inherently unfair.
The competition is not the goal - the point isn't to see "who's better", the competition with real humans is just a tool to help improve the reasoning, planning and decision making capabilities in a less artificial environment than usually. The purpose of that project is to use Starcraft as a playground to test, review and improve reinforcement learning methods so that they can be used for other needs.
Any aspect of "fairness" is relevant if and only if it helps that goal. APM limits are important because we want the computer to choose strategies that would actually be good choices in that situation even for a human player, instead of being able to win with strategies that are powerful only because of superhuman clicking speed. On the other hand, requiring the computer to press buttons with a robot hand doesn't really facilitate improvements to the decisionmaking part, so it's irrelevant even if it would make it more fair.
Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning
#105What would be interesting is to limit the AI processing speed to human capacity, which is something like 60 bits per second.
In all these AI v. Human games I see, it is really apples to oranges because the human consumes vastly less resources and compute cycles to perform at the same level as the AI. And when I say 'vast' I mean Vast. There is like a quintillion factor difference between the AI and the human.
There is no way the AI is even comparable to the human. To be comparable, we'd have to parallelize the game and lock millions of humans onto the game full time for millions of years.
At the end of the day, the AI is just a more sophisticated lookup table. There is as yet no analogous AI to human play.
Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning
#106Earlier quoted context omitted.
> Not being robust to strategies it hasn't seen before is a serious shortcoming in a real time strategy game. I'm not sure if you've actually played or followed competitive SC2. This is absolutely normal . Players will pull something completely unexpected out of a hat and win. The losing player will learn from it in future games. That's just how it goes. Unexpected strategies are really hard to counter when you've ne…
I don't think you understood his point. A player doesn't need to have seen a strategy before to react correctly to it. Sometimes a player will pull something completely unexpected out of a hat and lose, because the other guy reacted correctly thanks to his game experience. If you watch any high GM player stream half the time his reaction to what his (worse) opponent is doing is "WTF is this?" as he then proceeds to c…
I do understand his point. There are just as many examples of people being so overwhelmed by the new strategy that they don't know how to respond to it and lose. And then learn how to deal with it in later matches. Also, downvoting me for disagreeing is a dick move.
Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning
#107Earlier quoted context omitted.
If Google could tackle Civilizations god-aweful AI next, I'd be over the moon.
Do you mean Google should write a better AI than the one in Civilization? Interesting. But also consider this: in casual strategy videogames -- actually, strike "strategy" and just consider videogames -- most players don't want a really hard opponent. A computer opponent that is really very hard to beat is not what we want, because that'd be frustrating and many of us play videogames (yes, even strategy games!) to un…
In strategy games like Civilization the "preferred" difficulty level is that hard/easy to beat because it gets extra resources. It would be preferrable to have the same difficulty level through opponents that play better/smarter while having the same game mechanic consequences as players if they make the same actions, but we currently can't, so they get artificial production multipliers and such.
Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning
#108Earlier quoted context omitted.
for starters you just need a gps antenna on each soldier, an ammo count and a health meter (healthy, hit, safe). these 3 pieces of info already allow you to create a strategy framework
yes. and the soldiers need to do exactly what you tell them. good luck
this problem would happen also with a human generated strategy so i don't really understand what your point is.
food luck
Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning
#109Earlier quoted context omitted.
Wasn't blink stalker micro pretty central to its playstyle as protoss, the race it was best at? Why would you say that is minor and marginal?
It can no longer do micro like that, I have watched most of its games several times and it never did any micro feat like that again. On the contrary its micro was often lacking compared to humans, it won mostly through very strong timing attacks and uncanny ability to pick when to fight.
> It can no longer do micro like that
Is that supposed to be an intention in design? I'd figure an AI could easily outplay a human if that weren't the case, given its inherent advantages (e.g. better accuracy, the ability to instantaneously prioritize which units to blink in/out).
For example, [this](https://youtu.be/pETcAm82vXU?t=322) game is one example where an AI could perform even better using such tactics.
EDIT: another commenter mentioned it would make the game too unfair, so it was an intention in design.
Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning
#110So they added exploiter agents showing the AI what strats are on the meta. When the meta changes, will it still perform? Maybe im reading it wrong but seems much less impressive with the exploiter agents in the mix.