18-879N: Special Topics in Systems and Control: Multi-Agent Systems

Units: 12

Multi-agent systems (MAS's) occur in both the natural and the artificial world. Immune systems, nervous systems, multi-cellular organisms, ecologies, insect societies, distributed computing, communication networks, artificial neural networks, evolutionary algorithms, artificial life, economies, corporations and the internet are some examples. The distinguishing features of MASs are:

  1. Their decisions are made by a number of distinct entities, each with some degree of autonomy, and;
  2. Their behaviors are emergent (more complex than the mere sum of their individual behaviors) and often, counter-intuitive.

Many problems are best solved by MAS's, and most really large and difficult problems can only be solved by MAS's. The purpose of this course is to make participants aware of the state-of-the-art in MAS design, particularly, for software agents, and for combinations of software and human agents. The topics include natural exemplars of MAS's, and methods for task-decomposition, learning, competition, and cooperation among software agents.

Prerequisites: Graduate standing, or senior standing in CIT or Computer Science, or permission of the instructors.

Grading: There will be three-to-four short, group projects, and weekly progress reports. Grading will be on the basis of the presentations and final reports for the projects.

This course is cross-listed as 19-615.


Areas:

Signals and Systems, Artificial Intelligence, Robotics and Control

Designations:

Coverage
Last modified on 2006-10-23

Past semesters:

S07, S06, S05, S04