We are running the competition in 2023 at CoG 2023 (awaiting approval). The information below was for 2021.
Submission (Information on how to submit AIs developed with our OpenAI Gym API for ver. 4.40 or later is here (updated: July 8, 2019).)There are two deadlines : midterm and final. We also welcome submissions from non-students.
Please feel free to assess your AI's performance against other AIs by submitting it to us by the former deadline, based on which you can upgrade or tune your AI and resubmit it by the latter deadline.
Although in this year competition, three characters (Zen, Garnet, and Lud) are used, please send us only one AI developed in Java or Python that could play at any of the six leagues ( 3 characters x Standard or Speedrunning) by writing getCharacter() as follows:
@Override
public String getCharacter(){
return CHARACTER_ZEN; // #1
}
In #1, CHARACTER_GARNET, CHARACTER_LUD, and CHARACTER_KFM are also OK, but please write CHARACTER_ZEN as a default sentence. Please do not write something like
return "ZEN"; or return "CHARACTER_ZEN";
that cause errors.
For submission, please create a zip file containing
a) your AI's .jar (in case you use Java and since the competition address is gmail address, please change the extension ".jar" to, for example, "._jar"),
b) code (.java for a Java AI or .py for a Python AI) and/or
c) File-I/O folder with the same name as your AI (if any; in case you use machine learning mechanisms, you might need this folder for saving those trained data) as well as
d) a README file.
And please send the zip file to ftg.aic@gmail.com
with the mail subject "2021_FTG_AI". In the submission email, please
also attach a PowerPoint (or OpenOffice) file describing the following
information:
1. AI Name, Developers' Name(s) and Affiliation(s)
2. AI's Outline
The number of slides should be from 1 to 3 for midterm submission and 3
to 5 for final submission.
1 If you develop AIs in Python, please use Python 3.5, the official version of Python for this competition.
2 No deep learning libraries such as Torch or TensorFlow are supported at our competition servers. As a result, if your AI is a deep learning one, you should do as follows:
2.1 Submit an AI that does not need such deep learning libraries, but can perform forward propagation through the trained network from the input layer to the output layer and then conduct an action associated to one of the output units.
2.2 Store all trained weights in a folder with the same name as your AI. In the competition, this folder will be placed under ./aiData/
2.3 Refer to the following sample AIs which strictly follows 2.1 and 2.2:
LoadTorchWeightAI (in Java) and BasicBot (in Python) available in ------ this folder------
In addition, for Python deep learning AI developers, you might find the following site useful:
https://github.com/oreilly-japan/deep-learning-from-scratch
3 For fair competition among various approaches, use of GPUs is prohibited at the competition servers.
Please note that your AI's source codes and slides will be made publicly available after the final submission.
--------------------------------------------------------------------------------------------------------------------
#Midterm deadline (We recommend you do midterm submission, but if you miss it, you can still submit your AI to us by the final deadline.)
May 24, 2021 (23:59 JST)
#Notification of the midterm results
To be emailed to all submitters.
As written above, even though you might miss (might have missed) the mid-term submission, you can still submit your AI to us by the final deadline. However, we would appreciate it if you could send us an email to ftg.aic@gmail.com in advance indicating your intention of participation, so that we can keep you updated of any new information about the competition.
#Final deadline
July 29, 2021 (23:59 JST)(no extension!!)
#Notification of the final results
To be announced at CoG 2021's competition session and later at the results page here.
--------------------------------------------------------------------------------------------------------------------
Although we cannot take it for grant but, besides an IEEE winning certificate, money prizes might be provided by IEEE Student Games-Based Competition Comittee to the winning students. For your reference, in, say, the 2015 competition, IEEE provided money prizes (in total 1,800 USD) to qualified entries as follows:
Rank: 1 -> $500 USDAI: Machete
Members: Axel G. Garcia Krastek
Affiliation: Otto-von-Guericke University Magdeburg, Germany
Rank: 2 -> $300 USD
AI: Ni1mir4ri
Members: Jiyuu Yi
Affiliation: Sejong University, South Korea
Rank: 3 -> $500 USD (Top ranking student)
AI: Jay_Bot
Members: Manje Kim, Cheong-mok Bae
Affiliation: Sejong University, South Korea
Rank: 4 -> $300 USD (2nd ranking student)
AI: RatioBot
Members: Teerakit Vanitcharoennum, Natthawut Manchusoontornkul, Worawat Choensawat
Affiliation: Bangkok University, Thailand
Rank: 8- > $200 USD (3rd ranking student)
AI: GiantTeam
Members: Suwijak Wipachon, Thiti Rueangrit, Sutee Chamnankit, Kingkarn Sookhanaphibarn
Affiliation: Bangkok University, Thailand
Note that 1st to 3rd student ranking prizes went to top-three teams whose at least one student member attended CIG 2015.
--------------------------------------------------------------------------------------------------------------------