A Detailed Analysis of a Multi-agent Diverse Team

Abstract

Abstract.In an open system we can have many different kinds of agents. However, it is a challenge to decide which agents to pick when forming multi-agent teams. In some scenarios, agents coordinate by voting continuously. When forming such teams, should we focus on the diversity of the team or on the strength of each member? Can a team of diverse (and weak) agents outperform a uniform team of strong agents? We propose a new model to address these questions. Our key contributions include: (i) we show that a diverse team can overcome a uniform team and we give the necessary conditions for it to happen; (ii) we present optimal voting rules for a diverse team; (iii) we perform synthetic experiments that demonstrate that both diversity and strength contribute to the performance of a team; (iv) we show experiments that demonstrate the usefulness of our model in one of the most difficult challenges for Artificial Intelligence: Computer Go 1 .

The Power of Teams that Disagree: Team Formation in Large Action Spaces

“…We define a diverse team as a team of ST agents, and a uniform team as a team of NST agents. Therefore, our focus changes from modeling diverse teams as teams with different agents (as in models such as Marcolino, Jiang and Tambe (2013) [14]), to focusing on diverse teams as teams where the agents tend to disagree. This change allows us to make new predictions that were not possible before.…”

“…Recently, Marcolino, Jiang, and Tambe (2013) [14] showed the importance of diversity when forming teams of voting agents. They show that it is possible for a diverse team of weaker agents to overcome a uniform team of copies of the best agent, if the weaker agents are able to play better than the best agent at some world states.…”

“…By diversity, we mean agents that tend to disagree. In Marcolino, Jiang, and Tambe (2013) [14], a diverse team is defined as a set of agents with different pdfs. Hence, they disagree because of having different probabilities of playing certain actions.…”

“…Many researchers emphasize the importance of diversity when forming teams [11,10,8,14]. However, there are many important questions about diversity that were not asked before, and are not explored in such models.…”

“…Domains where the possible solutions are discrete, however, are not captured by such a model. Marcolino, Jiang, and Tambe (2013) [14] study teams of agents that vote in discrete solution spaces. They show that a diverse team of weaker agents can overcome a uniform team made of copies of the best agent.…”