In 2012, 2014, and 2015 successful workshops on AI for adversarial real-time games were held in response to the considerable interest in the subject and the limited time for reporting on the annual StarCraft competition in the main AIIDE conference. This year, we'd like to broaden the workshop scope a bit to cover AI for all kinds of strategy games in the hope to attract more submissions and to spark discussions between research groups focusing on board game, real-time strategy game, and general video game AI. The goal of this follow-up workshop is to again bring together AI researchers and game AI programmers from industry, who are interested in adversarial game AI, to present and exchange ideas on the subject, and to discuss how academia and game companies can work together to improve the state-of-the-art in AI for games.
The one-day workshop will consist of paper presentations on adversarial game related AI topics (listed below), game competition descriptions, a commented man-machine StarCraft match (most likely replays), perhaps an invited presentation, and a discussion on future research. The competition summary and results will be presented at the main AIIDE conference. Contributions will be peer-reviewed to meet AAAI workshop standards.
Contributions will be peer-reviewed to meet AAAI workshop standards.
This workshop welcomes original research contributions, position papers, software demos, and post-mortem game analyses in the area of AI for strategy games --- including FPS and RTS video games and turn based games. Topics include, but are not restricted to:
9:00am - 9:05am Welcome
9:05am - 9:35am Paper 1: StarCraft Winner Prediction
(Yaser Norouzzadeh)
Predict Winners in StarCraft based on time dependent
and time independent features
9:35am - 10:05am Paper 2: Improving Terrain Analysis and
Applications to RTS Game AI
(Alberto Uriarte)
Speeding up StarCraft map analysis from minutes to seconds
10:05am - 10:35am Paper 3: Experiments on Learning Action Probability
Models from Replay Data in RTS Games
(Santi Ontañón)
Biasing MCTS using learned policies in micro-RTS
10:35am - 11:00am Coffee Break
11:00am - 12:00pm Show and Tell, Part 1
Attendees presenting current work (5-10 min each).
Mainly paper presentation previews.
12:00pm - 1:30pm Lunch
1:30pm - 2:15pm Show and Tell, Part 2
2:15pm - 2:30pm StarCraft Competition (David Churchill)
More on this in the next session
2:30pm - 3:30pm Group Discussion, Part 1
New RTS game competition using Santi's micro-RTS system
3:30pm - 4:00pm Coffee Break
4:00pm - 5:00pm Group Discussion, Part 2
What's next in RTS game AI: deep networks vs./and search?
Learning policy networks for combat or the full game?
Learning state abstrations?
Search without forward model?
The following people have agreed to help organizing the workshop:
Other likely attendees (list compiled from research interest and
recent AIIDE attendance) are:
Hector Munoz-Avila, Johan Hagelb\"{a}ck, Risto Miikkulainen, Chris Darken,
Michael Mateas, Vadim Bulitko, Hendrik Baier, Andrew Budd, Martin Cerny,
Christopher Dragert, Alexander Jaffe, Owen Macindoe, Dana Nau, Eric Raboin,
Glen Robertson, Matthew Tretin, Sung Jun Park, Goncalo Pereira, Baylor Wetzel,
Marius Stanescu, the workshop PC, and other interested AIIDE 2016 attendees.