A New Approach for Feature Subset Selection using Quantum Inspired Owl Search Algorithm
A New Approach for Feature Subset Selection using Quantum Inspired Owl Search Algorithm
复制标题
DOI：
10.1109/icist49303.2020.9202140
复制
发表时间：
2020-09
期刊：
影响因子：
--
通讯作者：
A. K. Mandal;Rikta Sen;Saptarsi Goswami;A. Chakrabarti;B. Chakraborty
中科院分区：
文献类型：
--
作者：
A. K. Mandal;Rikta Sen;Saptarsi Goswami;A. Chakrabarti;B. Chakraborty
特征子集选择是通过剔除不相关和冗余的特征来为分类任务选择最优特征子集的方法。然而，由于其固有的指数时间复杂性，寻找最优特征子集是具有挑战性的。为了解决这个问题，元启发式算法经常被用来在合理的时间限制内找到次优解。提出了一种量子启发猫头鹰搜索算法(QIOSA)用于特征子集选择。在该方法中，特征被表示为量子叠加态，并使用量子旋转门来加速向最优特征集的搜索。在12个公开可用的基准数据集上进行了仿真实验，与作者提出的二进制猫头鹰搜索算法(BOSA)和其他基于种群的特征选择技术(包括二进制遗传算法(BGA)和二进制粒子群优化(BPSO))相比，该方法的效率得到了评估。实验结果表明，与其他元启发式算法相比，QIOSA算法提高了分类精度，有效地减少了特征个数。
Feature subset selection is the approach of selecting the optimal feature subset for the classification task by removing irrelevant and redundant features. However, searching for the optimal feature subset is challenging due to its inherent exponential time complexity. To address the problem, metaheuristics are frequently used for finding the sub-optimal solution in a reasonable time constraint. In this paper, a Quantum Inspired Owl Search Algorithm (QIOSA) for feature subset selection is proposed. In this method, features are represented as quantum superposition states, and quantum rotation gate is used to accelerate the search towards an optimal set of features. Simulation experiments have done to evaluate the efficiency of the proposed approach compared to Binary Owl Search Algorithm (BOSA) proposed earlier by authors and other population-based feature selection techniques, including Binary Genetic Algorithm (BGA) and Binary Particle Swarm optimization (BPSO) with twelve publicly available benchmark datasets. The experimental results show that QIOSA improves classification accuracy and effectively reduces the number of features compared to other metaheuristic algorithms.