Abstract:<p>Feature selection is a multi-objective optimization problem since it has two conflicting objectives: maximizing the classification accuracy and minimizing the number of the selected features. Due to the lack of selection pressures, most feature selection algorithms based on multi-objective optimization obtain many optimal solutions around the center of Pareto fronts. Penalty boundary interaction (PBI) decomposition approach provides fixed selection pressures for the population, but fixed selection pressures are hard to solve feature selection problems with complicated Pareto fronts. This paper proposes a novel feature selection algorithm based on multi-objective particle swarm optimization with adaptive strategies (MOPSO-ASFS) to improve selection pressures of the population. An adaptive penalty mechanism based on PBI parameter adjusts penalty values adaptively to enhance the selection pressures of the archive. An adaptive leading particle selection based on feature information combines the opposite mutation and the feature frequencies to improve the selection pressure of each particle. The proposed algorithm is compared with 6 related algorithms on 14 benchmark UCI datasets and 6 gene datasets. The experimental results show that MOPSO-ASFS can find optimal solutions with better convergence and diversity than comparison algorithms especially on the high dimensional datasets.</p>

What problem does this paper attempt to address?

The paper attempts to address the issue of how to improve the selection pressure of multi-objective optimization algorithms in feature selection to achieve better solution set diversity and convergence. Specifically, feature selection is a multi-objective optimization problem because it has two conflicting objectives: maximizing classification accuracy and minimizing the number of selected features. However, existing multi-objective optimization-based feature selection algorithms often find many optimal solutions near the center of the Pareto front, lacking sufficient selection pressure to address feature selection problems with complex Pareto fronts. To overcome this issue, the paper proposes a novel feature selection algorithm based on Multi-Objective Particle Swarm Optimization (MOPSO), called Multi-Objective Particle Swarm Optimization with Adaptive Strategy for Feature Selection (MOPSO-ASFS). The algorithm enhances the selection pressure of the population by introducing an adaptive penalty mechanism and an adaptive leader particle selection strategy, thereby improving the diversity and convergence of the solution set. The specific contributions include: 1. **Adaptive Penalty Mechanism**: An adaptive penalty mechanism based on the PBI parameter is incorporated into the archive update mechanism to maintain the diversity of the archive. The adaptive penalty value is dynamically adjusted according to the distance between the current solution and the weight vector, thereby enhancing the diversity of the archive. 2. **Adaptive Leader Particle Selection**: A novel adaptive leader particle selection strategy combining feature frequency and reverse mutation enhances the diversity of the population. Feature information reflects the search status of the current archive, and incorporating feature frequency and reverse mutation into leader particle selection can avoid repeated searches of known spaces and adaptively change the selection pressure of particles. 3. **Experimental Validation**: The proposed MOPSO-ASFS algorithm was compared with 6 related algorithms on 14 benchmark UCI datasets and 6 gene expression datasets. Experimental results show that MOPSO-ASFS exhibits better convergence and diversity on high-dimensional datasets. In summary, the paper aims to improve the performance of feature selection by enhancing the selection pressure of multi-objective optimization algorithms, especially when dealing with high-dimensional datasets.

Multi-objective particle swarm optimization with adaptive strategies for feature selection

A Feature Selection Method Based on Feature-Label Correlation Information and Self-Adaptive MOPSO

An improved feature selection method based on angle-guided multi-objective PSO and feature-label mutual information

Multi-Objective Self-Adaptive Particle Swarm Optimization for Large-Scale Feature Selection in Classification

Exploratory Study of Sorting Particle Swarm Optimizer for Multiobjective Design Optimization

Multiobjective Particle Swarm Optimization for Feature Selection With Fuzzy Cost

Multi-objective particle swarm optimization with R2 indicator and adaptive method

Feature selection using bare-bones particle swarm optimization with mutual information

Feature Selection Based on Adaptive Particle Swarm Optimization with Leadership Learning

A two-stage maintenance and multi-strategy selection for multi-objective particle swarm optimization

Adaptive multi-objective particle swarm optimization based on virtual Pareto front

An improved multiobjective particle swarm optimization algorithm based on tripartite competition mechanism

A particle swarm optimisation based multi-objective filter approach to feature selection for classification

A Multi-Objective Particle Swarm Optimisation for Filter-Based Feature Selection in Classification Problems

A multi-objective optimization algorithm for feature selection problems

Feature subset selection via an improved discretization-based particle swarm optimization

Multiobjective Particle Swarm Optimization Algorithm Based on Adaptive Angle Division.

A multi-objective particle swarm optimization algorithm based on two-archive mechanism

Many-objective optimization of feature selection based on two-level particle cooperation

MOCPSO: A multi-objective cooperative particle swarm optimization algorithm with dual search strategies

An evolutionary multi-objective optimization framework of discretization-based feature selection for classification