Amii is proud to celebrate Matthew (Matt) E. Taylor (Canada CIFAR AI Chair, Fellow-in-Residence at Amii and University of Alberta Associate Professor) for being elected General Co-Chair of the 2022 Adaptive Agents and Multi-Agents Systems (AAMAS) Conference!

As General Co-Chairs -- the top level of conference organization -- Taylor and Professor Catherine Pelachaud (Director of Research at the French National Centre for Scientific Research) will oversee and coordinate the conference, including assembling and tasking individuals to Chair different areas of the conference.

“I’ve been attending AAMAS regularly since 2005 --- it’s one of my favorite venues,” says Taylor. “Conferences are a great way for people across the world to learn what each other are doing, discuss new ideas, and form new collaborations. I’m really excited that I get to help the community by organizing this!”

In addition to this honour, Matt also won the Best Paper Award on the Blue Sky Ideas track at AAMAS 2021 with Diverse Auto-Curriculum is Critical for Successful Real-World Multiagent Learning Systems, the paper he co-wrote with Yaodong Yang, Jun Luo, Ying Wen, Oliver Slumbers, Daniel Graves, Haitham Bou Ammar and Jun Wang.

The paper explores the auto-curriculum framework and how agents in a multi-agent system can improve over time. Multiagent reinforcement learning (MARL) has achieved a remarkable amount of success in solving various types of video games. The authors argue that in order to extend MARL methods to real-world domains outside of video games, it is critical to maintain a diversity-aware auto-curriculum, allowing agents to train both in different contexts and while interacting with agents that have different preferences, motivations and capabilities.

We asked Taylor to explain this work at a high level:

"Suppose you learn to drive in Edmonton. If you fly to Los Angeles and start driving, not only do the roads look different -- why are there six lanes on the highway? Where is the snow? -- but other drivers’ behaviour is much more aggressive," he explained. "By training or 'practicing' in lots of different situations, you could be better prepared for when you experience novel situations."

"Winning this award, out of 43 papers, makes us even more excited to work on bringing multiagent learning out of the laboratory and into the real world."

• Matt Taylor, Amii Fellow-in-Residence

“We were very excited when this was announced in the closing session of AAMAS-21. Winning this award, out of 43 papers, makes us even more excited to work on bringing multiagent learning out of the laboratory and into the real world,” shares Taylor. “Our hope is that next year we won’t be talking about an abstract idea, but have an actual proof of concept!”

AAMAS is the largest and most influential conference in the area of agents and multiagent systems, bringing together researchers and practitioners in all areas of agent technology, as well as providing an internationally renowned, high-profile forum for publishing and finding out about the latest developments in the field. The conference includes a variety of activities, such as plenary keynote talks, parallel sessions with contributed talks, poster sessions, associated workshops, tutorials, and social events. AAMAS 2022 is scheduled to be held in Auckland, New Zealand.