查看文章

Memetic Multi-Agent System (MeMAS) emerges as an en-hanced version of multi-agent systems with the implementa-tion of meme-inspired agents. Previous research of MeMAS has developed a computational framework in which a series of memetic operations have been designed for implementing multiple interacting agents. This paper further endeavors to address the speci c challenges that arise in more com-plex multi-agent settings where agents share a common set-ting with other agents who have di erent and even compet-itive objectives. Particularly, we propose a memetic multi-agent demonstration learning approach (MeMAS-P) with improvement over existing work to allow agents to improve their performance by building candidate models and accord-ingly predicting behaviors of their opponents. Experiments based on an adapted mine eld navigation task have shown that MeMAS-P could provide agents with ability to acquire increasing level of learning capability and reduce the candi-date model space by sharing meme-inspired demonstrations with respect to their representative knowledge and unique candidate models.