A Spark-based Distributed Dragonfly Algorithm for Feature Selection

Dragonfly algorithm is an intelligent group optimization algorithm. In this paper, dragonfly algorithm is well used in the field of feature selection. However, dragonfly algorithm has the problem of falling into local optimal solution, which reduces the performance of feature selection and classification. Therefore, this paper proposes a binary Dragonfly optimization algorithm based on spark, which integrates the global optimization ability of dragonfly algorithm with the parallel computing ability of spark, greatly improving the performance of the algorithm. Experimental results show that binary Dragonfly algorithm has better performance than traditional particle swarm optimization algorithm and genetic algorithm. The binary Dragonfly algorithm based on spark solves the problem of falling into the local optimal solution, improves the running speed of the algorithm, can effectively deal with massive data, and greatly enhances the performance of the algorithm.