2012 aiide predictions?

https://github.com/bwapi/bwapi
krasi0
Posts: 254
Joined: Thu Dec 18, 2008 12:42 pm

Postby krasi0 » Wed Aug 08, 2012 5:48 am

[quote name='abcdefg' date='08 August 2012 - 02:28 AM' timestamp='1344382084' post='11037']
We just need to see some improved versions of the bots from previous years. I think if the bots worked on better engagements that would make very big difference but I know its easier said then done. A big battle with fog of war is not an easy feat to figure out things like how many siege tanks the enemy has and things like that.
[/quote]
Actually in an earlier version Skynet used to maintain positions of enemy units after they were hidden by FoW
abcdefg
Posts: 45
Joined: Sat Aug 20, 2011 2:37 am

Postby abcdefg » Thu Aug 09, 2012 11:16 pm

I hope some of the bot authors take the time to write brief papers just explaining their thoughts and reasoning about why they choose the techniques they did. That article on the overmind was interesting and helped spread the word to get more attention on why its worth trying to solve a problem like this. I really think the size and scope of a bot playing at a pro level is going to require a team of researchers like guys from IBM or something to really go at it like they did with Watson at jeopardy to help advance research a lot faster. I think overmind actually made a big contribution its just too bad they did not iterate on that bot another year it really could have been a force. They had a lot of cool and interesting ideas and it was just well executed although a little too reliant on mutas which admittedly are powerful.
Mass mutas is good but at pro level every single zerg does it so the metagame takes that into account and terran players know how to deal with it so again you cant just rely on 1 unit like mutas.
Kae
Posts: 35
Joined: Tue Jun 28, 2011 1:26 pm

Postby Kae » Fri Aug 10, 2012 1:23 am

[quote name='abcdefg' date='10 August 2012 - 08:16 AM' timestamp='1344554169' post='11040']
I hope some of the bot authors take the time to write brief papers just explaining their thoughts and reasoning about why they choose the techniques they did. That article on the overmind was interesting and helped spread the word to get more attention on why its worth trying to solve a problem like this. I really think the size and scope of a bot playing at a pro level is going to require a team of researchers like guys from IBM or something to really go at it like they did with Watson at jeopardy to help advance research a lot faster. I think overmind actually made a big contribution its just too bad they did not iterate on that bot another year it really could have been a force. They had a lot of cool and interesting ideas and it was just well executed although a little too reliant on mutas which admittedly are powerful.
Mass mutas is good but at pro level every single zerg does it so the metagame takes that into account and terran players know how to deal with it so again you cant just rely on 1 unit like mutas.
[/quote]

Actually, a lot of authors wrote scientific articles to explain some parts of their bot: UAlbertaBot, EISBot, BBQ, BTHAI, ... On the other hand, Overmind's team did not produce a single scientific article - I mean published in real scientific journals or conference proceedings, not an Ars Technica article.

Also, a survey showing the state-of-the-art is on progress, including some details on bots from authors who did not publish any papers yet on their AI, like Skynet or AIUR. About AIUR, I would love to publish something but I consider I have nothing interesting to present so far. I hope this survey to be published soon.

For curious people, it could be a good idea to store somewhere on broodwarai.com scientific papers about SC bots, if authors would agree to send pre-prints (I'm sure they will).

Return to “BWAPI (wrappers in other languages / questions and announcements related to BWAPI itself)”

Who is online

Users browsing this forum: No registered users and 1 guest