Elitist Multi-Objective Genetic Algorithm Form Optimization.

Date of Submission

December 2002

Date of Award

Winter 12-12-2003

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Master's Dissertation

Degree Name

Master of Technology

Subject Name

Computer Science


Machine Intelligence Unit (MIU-Kolkata)


Murthy, C. A. (MIU-Kolkata; ISI)

Abstract (Summary of the Work)

In many real world situations there may be several objectives that must be optimized simultaneously in order to solve a certain problem. The main difficulty in considering multi-objective optimization is that there is no accepted definition of optimum in this case, and therefore it is difficult to compare one solution with another. For example, the objactives of designing an engineering device can be its efficiency and cost involved, which are contradictory in general.One approach for solving multi-objective problems may be to optimize each cri- terion separately and combine the solutions thus obtained. However, this method is seldom likely to provide a solution where each criterion is optimally balanced. In fact, it may so happen that optimizing one objective may leacd to unacceptably low performance of another objective. Thus for solving multi-objective problems all the objectives need to be treated together. In general, these problems admit multiple solutions, each of which is considered acceptable and equivalent when the relative importance of the objectives is unknown. The best solution is subjective and depends on the need of the designer or decision maker.Genetic algorithm (GA)(6] is a stochastic process performing search over a com- plex and multidimensional space, in a randomized method in that it utilizes domain specific knowledge, in the form of objective function, to perform a directed random search. GAs involve a population based search in the solution space. Conventionally, GAs have been used for optimizing a single objective function. However, their popula-tion based nature make them conducive for extension to the multi-objective case as well.Traditional search and optimization methods such as gradient descent search, and other non-conventional ones such as simulated annealing(10] are difficult to extend to multi-objective case, since their basic design precludes such situations. For these meth- ods, the multi-objective problems have to be reformulated as single objective ones using appropriate combination techniques like weighting etc. On the contrary population based methods like evolutionary algorithms are well suited forhandling several criteria at the same time. In this work, we dea! with the issues involved in multi-objective opti- mization using genetic algorithms, and proposed ways of enhancing their performance.


ProQuest Collection ID: http://gateway.proquest.com/openurl?url_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:dissertation&res_dat=xri:pqm&rft_dat=xri:pqdiss:28843362

Control Number


Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.



This document is currently not available here.