But can your AI learn to drop mules so it can capture some escaping ultras?
Better link: https://clips.twitch.tv/BusyDrabFennelRedCoat
StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer
11–20 of 54 posts
Re: StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer
#12But can your AI learn to drop mules so it can capture some escaping ultras? Better link: https://clips.twitch.tv/BusyDrabFennelRedCoat
Funny enough, after I posted this as a joke I realized this is exactly the kind of thing an AI would be incredibly good at finding. It makes it all that much more impressive seeing Buyn do it.
Re: StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer
#13Interesting, outside of the academic merits of the research, I wonder if they could partner with Blizzard (or other similar RTS game studios) and start shipping games with AI that is very difficult to beat? That would usher in a new era of solo and offline game play that could be very fun and challenging! Plus we'd be training Skynet, err, I mean the neural network in the process!
My understanding is that a micromanagement-based AI is likely not to be an "interesting" difficult AI to beat. The DeepMind project is aimed more towards (with its apm caps) at building something that makes good decisions (e.g., unit composition, build orders, crisis management) instead of relying on brute force optimal control of units. While it is likely possible to completely break the balance of the game with perfect unit control, playing against such an AI (which is really just a bot at this point) would get stale pretty quickly.
Re: StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer
#14I've always felt like macro is pretty easy to automate, once you've identified a lot of the potentials in the early game you can get going pretty well. Is AI playing against each other getting to some sort of endgame consistently (against equal opponents), or are they having blowouts because of some small difference in midgame?
Also wondering what the interface is - are they creating a virtual mouse/keyboard to enter inputs, are they playing in some sort of headless mode? How do they deal with recognizing units?
Re: StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer
#15For those who don't know Starcraft: microing refers to the small scale/precise maneuvering of units during a battle, as well as activating special unit abilities at certain times. In the lower level leagues microing is not as important as having a good overall large-scale "macro" strategy. The microing really starts coming into play at the higher levels where it can give the edge needed to win a battle. I am not very…
How can this be so? In my experience playing Starcraft the decision space for macro is very, very, very limited. You are essentially locked into a handful of viable build orders for the early game, a handful of timing attack opportunities in the early and mid game, and after that things veer so off track that very few players do more than wing it.
I'd disagree. At the pro level, choices like unit composition and expansion timing are very strongly influenced by what the other player is doing. I've seen videos where a caster (like Lowko) breaks down "the build order" that a player did in a pro level game, only for the pro to turn around and say something like: I wasn't following a strict build order, just reacting to what I saw from the opponent.
I watched a pro game just the other day where one of the players noticed that their protoss opponent had a late timing on their second pylon in their main base - as in, it should have been there, but wasn't. That protoss player has a well-known penchant for early game stargate harassment, so the other player read this situation and reacted by delaying their first expansion and instead producing early-game air defenses.
As another example, take unit composition. If a terran player is running a bio-based army (marine, marauder, medivac), what sort of units should zerg produce? Infestors are usually a good choice in this matchup. On the other hand, mass infestors will be much worse against a primarily mech-based terran army; instead you might see many vipers on the field (which, vice versa, wouldn't be great against a bio army).
Pro players are essentially always scouting what their opponent's doing because the macro decisions can make or break the game. The general sentiment I've seen that is good macro skills will push you far up into SC2 leagues, and it's not until diamond/masters that micro starts to really matter.
Re: StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer
#16I've always felt like macro is pretty easy to automate, once you've identified a lot of the potentials in the early game you can get going pretty well. Is AI playing against each other getting to some sort of endgame consistently (against equal opponents), or are they having blowouts because of some small difference in midgame? Also wondering what the interface is - are they creating a virtual mouse/keyboard to enter…
If this project uses the BWAPI (which it says it does in the paper), then I assume that this doesn't need to recognize units as that state is available from the game. On the other hand, the DeepMind "research environment" (https://deepmind.com/blog/deepmind-and-blizzard-open-starcra...), presents a simplified view of the map that requires AIs to recognize units as simplified blobs on the map and issue commands that more closely resemble what a human would do via mouse and keyboard.
Re: StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer
#17Interesting, outside of the academic merits of the research, I wonder if they could partner with Blizzard (or other similar RTS game studios) and start shipping games with AI that is very difficult to beat? That would usher in a new era of solo and offline game play that could be very fun and challenging! Plus we'd be training Skynet, err, I mean the neural network in the process!
My understanding is that a micromanagement-based AI is likely not to be an "interesting" difficult AI to beat. The DeepMind project is aimed more towards (with its apm caps) at building something that makes good decisions (e.g., unit composition, build orders, crisis management) instead of relying on brute force optimal control of units. While it is likely possible to completely break the balance of the game with per…
Just like playing against MarineKing would get stale quickly for most players :P
The problem with starcraft is that its like chess - for a lot of players it really DOES get stale pretty quickly, with only one or two interesting moments in a game (assuming no early blunders). The games that are interesting involve near perfect play for 10-20 minutes and then frequently one split second decision. Macro changes or new strategies are quickly picked up and interated on, and I think thats what the hardcore still stick around for
Re: StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer
#18Earlier quoted context omitted.
My understanding is that a micromanagement-based AI is likely not to be an "interesting" difficult AI to beat. The DeepMind project is aimed more towards (with its apm caps) at building something that makes good decisions (e.g., unit composition, build orders, crisis management) instead of relying on brute force optimal control of units. While it is likely possible to completely break the balance of the game with per…
Just like playing against MarineKing would get stale quickly for most players :P The problem with starcraft is that its like chess - for a lot of players it really DOES get stale pretty quickly, with only one or two interesting moments in a game (assuming no early blunders). The games that are interesting involve near perfect play for 10-20 minutes and then frequently one split second decision. Macro changes or new s…
The difference is that an interesting AI could likely be scaled back meaningfully for lower skill levels. Make MarineKing play with only the mouse against a lower skilled opponent? The results might be less stale.
Rate limiting a brute-force AI would make it trivially bad.
Re: StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer
#19Earlier quoted context omitted.
How can this be so? In my experience playing Starcraft the decision space for macro is very, very, very limited. You are essentially locked into a handful of viable build orders for the early game, a handful of timing attack opportunities in the early and mid game, and after that things veer so off track that very few players do more than wing it.
I'd disagree. At the pro level, choices like unit composition and expansion timing are very strongly influenced by what the other player is doing. I've seen videos where a caster (like Lowko) breaks down "the build order" that a player did in a pro level game, only for the pro to turn around and say something like: I wasn't following a strict build order, just reacting to what I saw from the opponent. I watched a pro…
I completely agree with your assessment.
I watch a fair bit of pro play, and also play enough to bounce between low diamond and high platinum on NA. It is almost always my macro that wins or looses games.
Most people at my level wouldn't even have consistent enough play to be able to notice the difference between well and poorly micro'd games. But if they miss an upgrade, or build the wrong comp, or worse, don't expand, that is the win/loose for 98% of games below masters.
Which all around sucks for me, because my micro is on point :)
Re: StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer
#20For those who don't know Starcraft: microing refers to the small scale/precise maneuvering of units during a battle, as well as activating special unit abilities at certain times. In the lower level leagues microing is not as important as having a good overall large-scale "macro" strategy. The microing really starts coming into play at the higher levels where it can give the edge needed to win a battle. I am not very…
How can this be so? In my experience playing Starcraft the decision space for macro is very, very, very limited. You are essentially locked into a handful of viable build orders for the early game, a handful of timing attack opportunities in the early and mid game, and after that things veer so off track that very few players do more than wing it.
This is sometimes called the game's "meta" but it's very unclear how much of this is real and how much is basically just fashion.
Some good but not quite star quality players like Ketroc deliberately ignore the meta. Ketroc plays with loads of Ravens, he did this when people said Ravens were strong but hard to use, he still did it when the accepted wisdom was that they were now useless, and last time I looked he was still at it after yet further balance changes to SC2. How he uses them changes, but he likes Ravens and they never stopped being viable.
Games like Chess have meta too. A year ago humans believed that materiel was essential and strong position made up only for small differences like losing a pawn or two. Then Google's Alpha Zero AI says hold my beer, sacrificing loads of my pieces for positional advantage is fine, because I can use that advantage to pin your extra pieces and still mate you away.