Embedded gengtic algorithms for multiobjective optimization problem

This paper introduces a special class of genetic algorithm (GA) to solve a class of multiobjective optimization problems - the multiple objectives which are need to optimize cannot be expressed in terms of a single equation/weight. The design of an associative memory through cellular automata (CA) is a typical example of such type of problem. In this problem the two objectives: (i) finding out the structure of the attractor basins; and (ii) desired patterns sequence, cannot be related with each other by any equation. An efficient implementation of a new type of GA, termed as Embedded GA (EJGA) is used to solve this problem. The superiority of EGA over conventional GA and simulated annealing (SA) has been extensively established for CA based associative memory; thereby indicating that EGA is crucial for enhancing the performance of such multiobjective optimization problems.