Projection-Iterative-Methods-based Optimizer: A novel metaheuristic algorithm for continuous optimization problems and feature selection

doi:10.1016/j.knosys.2025.113978

Projection-Iterative-Methods-based Optimizer: A novel metaheuristic algorithm for continuous optimization problems and feature selection

2025-06-30

0

PRE

AI

DOI：10.1016/j.knosys.2025.113978

Original

Original request for help

Save

Abstract

En 中文

The Projection-Iterative-Methods-based Optimizer (PIMO) is a novel metaheuristic algorithm inspired by projection iterative methods. PIMO introduces four new operators to guide the population towards optimal convergence while enhancing both exploration and convergence speed. This approach, presented for the first time, integrates techniques such as Kaczmarz and stochastic gradient descent to improve performance and prevent convergence to local optima. The effectiveness of PIMO is validated through three sets of experiments: the CEC2017 benchmark functions, four real-world constrained problems, and twelve UCI datasets, where it outperforms five excellent feature selection algorithms. The numerical results indicate that PIMO consistently surpasses other algorithms across various problems, including eleven highly referenced new algorithms, seven state-of-the-art algorithms, seven novel mathematics-inspired algorithms, and five leading binary algorithms. The findings confirm that PIMO is robust, user-friendly, and effective for both continuous and discrete problem-solving. The source code for the PIMO is publicly available at https://www.mathworks.com/matlabcentral/fileexchange/181013-pimo .

Keywords:

Projection-Iterative-Methods

Metaheuristic Algorithm

Optimization

Feature Selection

CEC2017 Benchmark

Projection-Iterative-Methods-based Optimizer: A novel metaheuristic algorithm for continuous optimization problems and feature selection

Abstract

Journal

Organization