Biogeography-Based Optimization for Noisy Fitness Functions

 

Haiping Ma, Minrui Fei, Dan Simon, and Mei Yu

 

Biogeography-based optimization (BBO) is an evolutionary optimization algorithm. Noisy fitness functions are fitness functions that have noise superimposed on them. That means that when we evaluate how good a candidate solution is for an optimization problem, the result of the evaluation is noisy. The research discussed on this web page is an extension of BBO for noisy fitness functions. Noisy fitness functions arise in many practical applications. In fact, almost all real-world optimization problems probably include noise when we evaluate candidate solutions.

 

The paper below proposes a resampling-based BBO algorithm for noisy fitness functions, applies it to benchmark problems, and compares it with Kalman filter-based BBO, which is another BBO extension for noisy fitness functions. The paper also compares resampling-based BBO with resampling-based differential evolution (DE) and resampling-based particle swarm optimization (PSO). The MATLAB software that was used to derive the results in the paper can be downloaded in a zip file (start with the “readme.txt” file).

 

Reference

 

Haiping Ma, Minrui Fei, Dan Simon, and Mei Yu, “Biogeography-Based Optimization in Noisy Environments,” submitted for publication - pdf, 365 KB

 


Professor Simon’s Home Page

 

Department of Electrical and Computer Engineering

 

Cleveland State University

 


Last Revised: October 24, 2012