Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Grandmaster level in StarCraft II using multi-agent reinforcement learning

deepmind.com

Post reply on HN
Print view View original

31–40 of 321 posts

Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning

#31

Post by username90 »

hoseja wrote: post #15

There 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.

They are testing the AI's ability to do tactics and strategy, not motor and visual tasks. It has built in delay and needs to move around the map to gather information just like a human, so it doesn't have any significant unfair advantages.

Edit: Note that the version they sent out to the ladder had significant larger delay, significant lower APM, and didn't get any information not visible on the screen unlike the first iteration.

Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning

#32

Post by xrisk »

sailingparrot wrote: post #23
xrisk wrote: post #8

Earlier quoted context omitted.

Isn't this trained via playing against itself? I don't think there's a dataset involved.

AlphaStar is first trained in a supervised way using the publicly available replays from SCII matches. Even if you just consider self-play, I don't see how you would create a simulation that is realistic enough so that you could generalize afterwards to reality. RL today has an extremely hard time dealing with different distributions. You can see the amount of work OpenAI [1] had to do to go from a virtual hand+rubik…

Ah, ok. I've read about OpenAI Five (it plays Dota 2); and as far as I understand it's fully trained with self-play.

Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning

#33

Post by v4dok »

Iv wrote: post #12
nightcracker wrote: post #5

> After 50 games, however, DeepMind hit a snag. Some players had noticed that three user accounts on the Battle.net gaming platform had played the exact same number of StarCraft II games over a similar time frame — the three accounts that AlphaStar was secretly using. When watching replays of these matches, players noticed that the account owner was performing actions that would be extremely difficult, if not impossi…

To play against alphastar, you have to opt-in. IIRC it states Alphastar will hide itself. One of the things people noticed in replays was the lack of control groups and in the case of zergs, the ability to select larvas directly, which no player ever does. It could have been as simple as removing these quirks.

Nobody selects larvas directly? I guess I am a computer then. Way worse than Alphastar tho.

Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning

#34

Post by SiempreViernes »

mc32 wrote: post #4

How soon is this put into the hands of battlefield commanders? Not presuming it’s a bad thing, but I wonder when we’ll hear the outcome of a battle was aided by this kind of intelligence.

Oh god, I feel another 50 articles with AI in the title incoming to War on The Rocks...

Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning

#38

Post by bjornjaja »

hoseja wrote: post #15

There 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.

Interesting thing about AI is no matter how you spin things, the more advanced it becomes, unfairness grows proportionally.

After all, isn’t that the point of AI? To perform better at certain tasks than humans? We are visitors in the digital realm, just like when we put on a scuba suit and jump in the ocean—even the smallest fish can out-swim us.

Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning

#39

Post by Nimitz14 »

nightcracker wrote: post #5

> After 50 games, however, DeepMind hit a snag. Some players had noticed that three user accounts on the Battle.net gaming platform had played the exact same number of StarCraft II games over a similar time frame — the three accounts that AlphaStar was secretly using. When watching replays of these matches, players noticed that the account owner was performing actions that would be extremely difficult, if not impossi…

Funny how they put it as:

> When watching replays of these matches, players noticed that the account owner was performing actions that would be extremely difficult, if not impossible, for a human.

framing it like the agent was doing some sort insane play, when the reality is the main evidence of a player being alphastar (other than its garbage decision making) was the fact that it wasn't using hotkeys!

Re: Grandmaster level in StarCraft II using multi-agent reinforcement learning

#40

Post by smolder »

Iv wrote: post #12
nightcracker wrote: post #5

> After 50 games, however, DeepMind hit a snag. Some players had noticed that three user accounts on the Battle.net gaming platform had played the exact same number of StarCraft II games over a similar time frame — the three accounts that AlphaStar was secretly using. When watching replays of these matches, players noticed that the account owner was performing actions that would be extremely difficult, if not impossi…

To play against alphastar, you have to opt-in. IIRC it states Alphastar will hide itself. One of the things people noticed in replays was the lack of control groups and in the case of zergs, the ability to select larvas directly, which no player ever does. It could have been as simple as removing these quirks.

New players (like when I first played) are likely to select individual larva before they do tutorials or learn the hotkeys, but you're right that high level players would almost never select one directly. Maybe after a hatchery has died and there are still larva remaining?
Post reply on HN