EECS 372/472:
Designing & Constructing Models With Multi-Agent Languages


Home
Course Description
Contact Info.
Handouts
Student Projects
Related Links
Syllabus

Course Description

Brief Course Summary

This course focuses on the exploration, construction and analysis of multi-agent models, which simulate "emergent" scientific phenomena in a wide variety of content domains (such as birds flocking, diseases spreading, or consumers affecting market dynamics). We use state of the art agent-based modeling and "complexity science" tools, and discuss methodology for replication, verification and validation of models.

Curricular Requirement Satisfaction

  • For CS and CIS undergraduate majors, it counts towards:
    one credit of the project course requirement
    the A.I. or Interfaces area breadth and depth requirements
  • For Cognitive Science undergraduate majors, it counts as:
    an advanced elective credit for the Cognitive Science major.
  • For Learning Sciences graduate students, it fulfills:
    a design course requirement
  • For graduate students working the Cognitive Science Specialization: it counts as:
    one course credit toward specialization

Since agent-based modeling is a useful technique in numerous disciplines, we believe that students from a broad spectrum of backgrounds can benefit from this course, and we encourage diversity in enrollment, which is why there are no pre-requisites for this course. In addition to computer/cognitive/learning scientists, we welcome engineers, economists, linguists, biologists, mathematicians, industrial engineering and management scientists, business students, and any other students interested in social networks, distributed systems, or other phenomena with many interacting elements following similar rules.

Outline of Topics Covered

  • What is an agent?
  • Stationary and movable agents
  • Interactions between agents
  • Agent topologies
  • Properties of networks
  • Applications of ABM
  • Artificial Life
  • Comparison with Systems Dynamics Models
  • Integration of Machine Learning
  • Evolutionary computation
  • Systematic exploration of model parameter space
  • Verification of model specification
  • Replication of models
  • Validation of models
  • Connecting ABM with physical devices
  • Sensors and motors
  • Combining human and virtual agents
  • Participatory simulations

Course Objectives for students:

  • When a student completes this course, he/she should be able to:
    Identify core mechanisms of novel agent-based models
    Identify trade-offs in the design and use of agent-based topologies
    Construct original multi-agent Use behavior run and analysis tools to analyze model parameter space
    Verify and validate agent-based models
    Apply agent-based modeling to both scientific and everyday phenomena

Grading

  • Class-participation: 20%
  • Homework Assignments: 30%
  • Final Project: 50%
  • There will be no exams for this course.
  • EECS 472 enrollees will have additional coursework