Development of chaotic maps embedded particle swarm optimization algorithms

In this thesis, addition and modifications to particle swarm optimization (PSO) algorithm which is an optimization technique developed inspiring by movement of flock of birds have been performed. Especially, chaos which has been regarded as one of the soft computing techniques has been embedded to PSO in order to increase the convergence speed by escaping from the local optimum points; and twelve PSO algorithms with the name ?chaos embedded PSO algorithms? have been proposed. Performance comparisons of these algorithms with the other PSO algorithms which have been reported to have good performance in the literature have been performed. Furthermore, it has been shown that interval algebra which is a branch of rough set, regarded as one of the soft computing techniques, can be effectively used with PSO for problems in which continuous decision variables and intervals should be used as representation, and PSO computation related to this representation have been described. ?Rough PSO? has been proposed for this purpose and has been shown to be effectively used in rule mining within continuous valued variables as a first application. PSO, chaos, and rough set have been combined and general purposed PSO algorithms with the name ?rough chaotic PSO algorithms? have been proposed. These algorithms have been firstly used in numeric association rules mining in which there is not an efficient and automatic technique. Promising results have been obtained. Various modifications have been performed for the PSO to let it work for multiobjective optimization problems and first application has been performed in classification rule mining task of data mining. Efficient results according to the objectives have been obtained. Lastly, new PSO algorithms, ?multi-objective rough chaotic PSO algorithms?, which include multi-objective PSO, chaos, and rough sets combination, have been proposed and have been applied in data mining in order to find efficient solutions. Keywords: Particle swarm optimization, chaotic maps, interval algebra, multi-objective optimization, numeric association rule mining, classification rule mining.