Publication - Towards Bounded-Rationality in Multi-Agent Systems: A Reinforcement Learning Based Approach

Authors: Raja, Anita, and Lesser, Victor
Title: Towards Bounded-Rationality in Multi-Agent Systems: A Reinforcement Learning Based Approach
Abstract: Sophisticated agents operating in open environments must make complex real-time control decisions on scheduling and coordination of domain activities. These decisions are made in the context of limited resources and uncertainty about outcomes of activities. Many efficient architectures and algorithms that support these control activities have been developed and studied. However, none of these architectures explicitly reason about the consumption of time and other resources by control activities, which may degrade an agentís performance. The question of how to sequence domain and control activities without consuming too many resources in the process, is the meta-level control problem for a resource-bounded rational agent. The focus of this research is to provide effective allocation of computation and improved performance of individual agents in a cooperative multi-agent system. This is done by approximating the ideal solution to meta-level decisions made by these agents using reinforcement learning methods. Our approach is to design and build a meta-level control framework with bounded computational overhead. This framework will support decisions on when to accept, delay or reject a new task, when it is appropriate to negotiate with another agent, whether to renegotiate when a negotiation task fails and how much effort to put into scheduling when reasoning about a new task. The major contributions of this work will be: a resource-bounded framework that supports detailed reasoning about scheduling and coordination costs; a scheduling paradigm that can support parameters to control scheduling effort, horizon and slack; and a simulation environment for testing and comparing the performance of agents.
Keywords: Multi-Agent Systems, Negotiation, Planning, Scheduling
Publication: University of Massachusetts Computer Science Technical Report, Vol: 2001, Num: 34
Date: August 2001
Sources: PS: ftp://ftp.cs.umass.edu/pub/techrept/techreport/2001/UM-CS-2001-034.ps
PDF: /Documents/UM-CS-2001-034.pdf
Reference: Raja, Anita, and Lesser, Victor. Towards Bounded-Rationality in Multi-Agent Systems: A Reinforcement Learning Based Approach. University of Massachusetts Computer Science Technical Report, Volume 2001, Number 34. August 2001.
bibtex:
@article{Raja-215,
  author    = "Anita Raja and Victor Lesser",
  title     = "{Towards Bounded-Rationality in Multi-Agent
               Systems: A Reinforcement Learning  Based Approach}",
  journal   = "University of Massachusetts Computer Science
               Technical Report",
  volume    = "2001",
  number    = "34",
  month     = "August",
  year      = "2001",
  url       = "http://mas.cs.umass.edu/paper/215",
}