Abstract

"This thesis presents a new class of evolutionary algorithms called mobile cellular evolutionary algorithms (mcEAs). These algorithms are characterized by individuals moving around on a spatial population structure. As a primary objective, this thesis aims to show that by controlling the population density and mobility in mcEAs, it is possible to achieve much better control over the rate of convergence than what is already possible in existing cellular EAs. Using the observations and results from this investigation into selection pressure in mcEAs, a general architecture for developing agent-based evolutionary algorithms called Artificial Ecosystems (AES) is presented. A simple agent-based EA is developed within the scope of AES is presented with two individual-based bottom-up schemes to achieve dynamic population sizing. Experiments with a test suite of optimization problems show that both mcEAs and the agent-based EA produced results comparable to the best solutions found by cellular EAs"--Abstract, page iii.

Advisor(s)

Agarwal, Sanjeev, 1971-
Madria, Sanjay Kumar

Committee Member(s)

Tauritz, Daniel R.

Department(s)

Computer Science

Degree Name

M.S. in Computer Science

Publisher

Missouri University of Science and Technology

Publication Date

Summer 2010

Pagination

ix, 61 pages

Note about bibliography

Includes bibliographical references (pages 74-75).

Rights

© 2010 Srinivasa Shivakar Vulli, All rights reserved.

Document Type

Thesis - Open Access

File Type

text

Language

English

Library of Congress Subject Headings

Closed ecological systems -- Mathematical models
Computer algorithms -- Design
Evolutionary computation
Evolutionary programming (Computer science)
Multiagent systems

Thesis Number

T 9689

Print OCLC #

689997058

Electronic OCLC #

648786130

Share

 
COinS