Abstract: Data mining is an extraction of knowledge from large data set. It is an evolving technology which is a direct result of the increasing use of computer databases in order to store and retrieve information effectively. Optimization is essential for a huge amount of data processing. So that optimization is a challenging issue in data mining. Ant colony optimization (ACO) is an evolutionary computation technique. Proposed methodology is the iterative process of ACO algorithm, inertia weight adjustment is usually expected to make particles have stronger global searching capability in early stage to prevent premature convergence and have stronger local search capability in latter stage to accelerate convergent speed. In other words, the inertia weight should vary nonlinearly along with the process of decreasing slowly, then rapid and then slowly again so as to attain fast convergence speed in prophase and have local search capability to a certain degree at the later stage, too. In this dissertation ACO based k-means clustering is applied to generate clusters. And provide multimodal and higher dimensional complicated optimization problems, and can accelerate convergence speed, improve optimization quality effectively in comparison to the algorithms of ACO K-means.

Keywords: data Mining, Clustering, Ant colony optimization, K-means.