2021/07/29 by Hritam Basak, Basak, Hritam, Mayukhmali Das +3
Computer Science · Engineering · #Advanced Multi-Objective Optimization Algorithms #Artificial Immune Systems Applications #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Metaheuristic Optimization Algorithms Research
paper · pdf · doi:10.48550/arxiv.2107.14199
openalex publication_date 2021/07/29 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Swarm optimization algorithms are widely used for feature selection before\ndata mining and machine learning applications. The metaheuristic\nnature-inspired feature selection approaches are used for single-objective\noptimization tasks, though the major problem is their frequent premature\nconvergence, leading to weak contribution to data mining. In this paper, we\npropose a novel feature selection algorithm named Reinforced Swarm Optimization\n(RSO) leveraging some of the existing problems in feature selection. This\nalgorithm embeds the widely used Bee Swarm Optimization (BSO) algorithm along\nwith Reinforcement Learning (RL) to maximize the reward of a superior search\nagent and punish the inferior ones. This hybrid optimization algorithm is more\nadaptive and robust with a good balance between exploitation and exploration of\nthe search space. The proposed method is evaluated on 25 widely known UCI\ndatasets containing a perfect blend of balanced and imbalanced data. The\nobtained results are compared with several other popular and recent feature\nselection algorithms with similar classifier configurations. The experimental\noutcome shows that our proposed model outperforms BSO in 22 out of 25 instances\n(88%). Moreover, experimental results also show that RSO performs the best\namong all the methods compared in this paper in 19 out of 25 cases (76%),\nestablishing the superiority of our proposed method.\n