A Hybrid Particle Swarm Optimization with Binary Ali Baba and the Forty Thieves Algorithm for Feature Selection

The recent progress in technology enables a massive collection of databases. Noisy, redundant features significantly affect the machine-learning model's performance. The researchers adopted feature selection methods to solve this unavoidable scenario. It not only helps to deplete the dimension of the dataset but also plays a pivotal role in improving classification accuracy. This study presents a novel hybrid particle swarm optimization with Binary Ali Baba and the forty thieves' algorithm (BAFT) to select the significant features. The proposed model hybridizes the binary AFT, which provides a suitable search strategy with classical particle swarm optimization, as it is known for its convergence capability toward the best global solutions. KNN is used as a classifier to evaluate the efficacy of the proposed model. Using 07 different datasets, the performance of the proposed model is evaluated and compared with the traditional PSO, BAFT algorithm. To avoid the issue of overfitting, the dataset is subdivided into training and testing datasets with a 10-fold CV. The computational results demonstrate that the proposed method is more efficient than current state-of-the-art techniques.