Abstract

Individual-Based modeling has gained popularity over the last decade, mainly due to its proven ability to address a variety of problems, including modeling complex systems from bottom-up, providing relationships between component level and system level parameters, and relating emergent system level behaviors from simple component level interactions. Availability of computational power to run simulation models with thousands to millions of agents is another driving force in the wide-spread adoption of individual-Based modeling. in this paper, we propose an individual-Based modeling approach to solve engineering design and optimization problems using artificial ecosystems (AES). the problem to be solved is "mapped" to an appropriate AES consisting of an environment and one or more evolving species. the AES is then allowed to evolve. the optimal solution emerges through the interactions of individuals amongst themselves and their environment. the fitness function or selection mechanism is internal to the ecosystem and is based on the interactions between individuals, which makes the proposed approach attractive for design and optimization in complex systems, where formulation of a global fitness function is often complicated. the efficacy of the proposed approach is demonstrated using the problem of parameter estimation for binary texture synthesis. Copyright 2008 ACM.

Department(s)

Electrical and Computer Engineering

Keywords and Phrases

Artificial ecosystem; Individual-based modeling; Markov random fields; Optimization; Parameter estimation

International Standard Book Number (ISBN)

978-160558130-9

Document Type

Article - Conference proceedings

Document Version

Citation

File Type

text

Language(s)

English

Rights

© 2024 Association for Computing Machinery, All rights reserved.

Publication Date

01 Jan 2008

Download

Full Text Link

Share

 
COinS