An Improved Chaotic ACO Clustering Algorithm
Lei Yang, Kangshun Li, Wensheng Zhang, Zhenxu Ke, Kehui Xiao, Zhiguo Du · 2018
Based on the traditional Ant Colony Optimization (ACO) clustering algorithm (ACOC), this paper proposed a new ACO algorithm for clustering, named ACO-Based Clustering Algorithm with Chaos (ACOCC). The key improvement of this algorithm is the application of chaotic function to the initialization phase and updating phase of pheromone. Applying the chaotic function to the initialization phase of pheromone can enable ants to be distributed in different states as much as possible, while applying the chaotic function to the updating phase of pheromone can bring chaotic disturbance into the algorithm, which makes the ants to explore as many paths as possible and avoids premature and converging to suboptimal solutions. The improved algorithm was implemented by MATLAB and was experimented on several public datasets, our numerical results show that the improved algorithm ACOCC can produce better optimal solution, achieve the optimal solution faster, and produce optimal solutions stably.