We hold a competition of automated negotiation agents as a workshop in PRIMA (Principles and Practice of Multi-Agent Systems). Automated negotiation is one of the most important AI technologies which promote collaboration and orchestration among AI-enabled social entities. To populate the technology, automated negotiating agents competition (ANAC) has been held since 2010 in conjunction with the International Joint Conference on Autonomous Agents and Multi-Agent Systems (AAMAS) or the International Joint Conference on Artificial Intelligence (IJCAI). With ANAC committee, we hold the first satellite competition in PRIMA. We welcome your submission of AI agents as well as your participation and discussion in the workshop.
Today’s remarkable AI trends may accelerate the necessity of collaboration among AIs and game theoretic AI technologies. Among other thing, negotiation is quite practical and suitable to apply to today’s real IT systems. That is because the protocols used in negotiation is quite similar to a real business transaction, which consists of request for proposal, proposal, counter proposal, and so on.
On the other hand, the most invested, advanced, and advertised match-type AI is no more than game; Go, Quiz, and Video Game. Even though researches on these games have been accelerating fundamental AI researches, straightforward implementations in real business seemed to be hard. Our approach to this problem is to set up a business oriented competition first, which can be easily transported into real business, then gather AI researchers to develop an eco-system.
Therefore, in addition to hold a competition of standard automated negotiation (bilateral negotiation with alternative proposals), we will introduce two demonstrations of the setup of more business-like setting of competitions (leagues) that will be part of ANAC 2019 and hold an exhibition of the game setting in PRIANAC. One is chained-negotiation for supply chain management in the context of Industrie 4.0. The other is negotiation with an uncertain utility function to capture the preference of human.
4 hours ( 1:00 PM - 5:00 PM)
Challenge: agents can make use of bids in previous negotiation sessions. Therefore, we explicitly permit the use of local file and a popular machine learning library (scikit-learn).
A submission includes the following files. All following files should be compressed in a zip file.:
By entering the competition, the participants agree on granting copyright to our organisation committee to make source code and explanations of the agents available on the website of the competition and repository of agents with the open source GENIUS, while retaining copyright yourselves as well.
The registration of workshop is necessary from the PRIMA registration site.
It is expected that teams that make it through to the finals will have a representative attending PRIANAC session. Teams in the final will be given the opportunity to give a brief presentation describing their agent at the session. It is an option to present through Skype or a video message without attending PRIANAC (no registration fee). Those who would like to make a presentation and are not a finalist, please contact us.
AIST: National Institute of Advanced Industrial Science and Technology