AAMAS-04 Workshop
on
Learning and Evolution in Agent Based Systems
July 20, 2004
New York

Organizing committee members:

• Sandip Sen (Chair), Univ. of Tulsa, sandip@kolkata.mcs.utulsa.edu
• Eugenio Oliveira, Faculdade de Engenharia Universidade do Porto, eco at fe.up.pt
• Enric Plaza, IIIA, Artificial Intelligence Research Institute, enric at iiia.csic.es
• Peter Stone, University of Texas, Austin, pstone at cs.utexas.edu

• Stephane Airiau, Nilanjan Ganguly, Sandip Sen, Sabyasachi Saha, "Evolutionary tournament-based comparison of learning and non-learning strategies for iterated games,"
• Teddy Candale & Sandip Sen, "Choosing satisficing service providers by learning referrals"
• Jacob Crandall and Michael A. Goodrich, "Establishing Reputation Using Social Committment in Repeated Games"
• Partha S. Dutta, Srinandan Dasmahapatra, Steve R. Gunn, Nicholas R. Jennings, and Luc Moreau, "Strategic Communication to Improve Distributed Learning"
• Vanessa Frias-Martinez and Elizabeth Sklar, "A team-based co-evolutionary approach to multi agent learning"
• Matthew E. Gaston, John Simmons, and Marie DesJardines, "Learning to Form Teams in Networks"
• Yoav Horman and Gal A. Kaminka, "Improving Sequence Learning for Modeling Other Agents"
• Sebastien Paquet, Nicolas Bernier and Brahim Chaib-draa, "Selective Perception Learning for Tasks Allocation"
• Avi Rosenfeld, Gal A Kaminka, and Sarit Kraus, "Adaptive Robot Coordination using Interference Metrics"

• Myriam Abramson and Ranjeev Mittu, "Joint Intentions and Joint Actions: Closing the Loop"
• Xin Li, Qingping Tao, and Leen-Kiat Soh, "Learning Negotiation Approach Selection with SVM"

Researchers in machine learning and adaptive systems have been addressing issues concerned with learning and adapting from past experience, observation, failures, etc. Whereas most of this research have focused on techniques for acquisition and effective use of problem solving knowledge from the viewpoint of a single autonomous agent, recent investigations have opened the possibility of application of some of these techniques in multiagent settings. The goal of this workshop is to focus on research that will address unique requirements for agents learning and adapting to their environment. Recognizing the applicability and limitations of current machine learning research when applied to situated agents will be of particular relevance to this workshop. The workshop will also encourage presentation and discussion of ideas relating to evolutionary learning and adaptation techniques in the context of agent based systems. Expected contributions include both evolutionary learning by individual agents as well as evolutionary design of agent societies. Of particular interest is new models for coevolving agent populations. We solicit research contributions that address new learning modalities, e.g., the use of communication to enhance learning. Presentation of applications of learning in key multiagent problems like negotiation, teamwork, trust, auctions, supply chains, etc. are welcome. We focus on three different ways in which machine learning can be used to enhance the performance of an Agent Based System:

1. An agent can learn the preferences and changing priorities of associated users.
2. An agent can learn about other agents in the environment in order to compete and/or cooperate with them. An agent can learn from other agents, taking advantage of their experiences and incorporating these into its own knowledge base.
3. An agent can learn about other regularities in its environment.

• Evaluation of the effectiveness of individual learning strategies (e.g., case-based, explanation-based, inductive, reinforcement), or multistrategy combinations.
• Characterization of learning and adaptation methods in terms of modeling power, communication abilities, knowledge requirement, processing abilities of individual agents.
• Developing learning and adaptation strategies, or reward structures, for environments with cooperative agents, selfish agents, partially cooperative (will cooperate only if individual goals are not sacrificed) and for environments that can contain mixture of these types of agents.
• Analyzing convergence properties of existing algorithms and constructing algorithms that guarantee convergence and stability of group behavior.
• Evaluating effects of knowledge acquisition mechanisms on responsiveness of agents or groups to changes in the agent population in the environment.
• Learning to work as an effective team by taking advantage of complementary skills and resources.
• Agents learning via passive or non-intrusive observation of user behaviors or by mimicking other agents.
• Evolving agent behaviors or co-evolving multiple agents with similar/opposing interests.
• Investigation of teacher-student relationships between agents or between an agent and the associated user.
• Applications of learning agents including agents that learn to negotiate contracts, learning trustworthiness of other agents, learn to detect security threats, etc.

E-mail the URL of either

• a brief statement of interest (1 page), or
• a complete paper (3000 words maximum) including keywords, word count, and authors' complete address

• Intent to submit: April 9. 2004
  
• Deadline for paper submission: Due to administrative requirements we strongly recommend submission by April 12, 2004 though we will consider submissions, with prior approval, until April 19, 2004
  
• Acceptance notice to participants: May 1, 2004
  
• Camera-ready papers due: May 24, 2004

Sandip Sen's homepage