Association Rule Mining by Dynamic Neighborhood Selection in Particle Swarm Optimization

Provided by: International Journals of Advanced Information Science and Technology (IJAIST)
Topic: Security
Format: PDF
Association Rule (AR) mining is one of the most studied tasks in data mining community with focus on improving computational efficiency. The standard Particle Swarm Optimization (PSO) is an evolutionary algorithm originally developed to simulate the behavior of birds and successfully applied for mining association rules. The problem with Particle swarm optimization algorithm is its trapping into local optima. This result in premature convergence of the algorithm affecting the efficiency of the rules mined. To improve the performance of PSO and maintain diversity of particles, a dynamic neighborhood selection in PSO is proposed for mining ARs. Dynamic neighborhood selection in PSO introduces the concept of local Best particle (lBest) replacing the particle best (pbest).

Find By Topic