Human Personality-Based Particle Swarm Optimization: A Trait-Driven Framework for High-Dimensional Heterogeneous Problems

This paper introduces a novel optimization algorithm, Human Personality-Based Particle Swarm Optimization (HPPSO), which integrates psychological modeling into the particle swarm optimization framework. Unlike conventional approaches that treat particles as indistinct units, HPPSO models each particle as an agent endowed with individualized human-like personality traits—such as curiosity, conformity, aggressiveness, risk-taking, and social behavior. These traits are mathematically formulated and embedded into the velocity and position update equations, enabling adaptive and diverse behavioral dynamics that enhance the exploration–exploitation balance. By explicitly introducing psychological diversity into swarm behavior, HPPSO opens a new research direction at the intersection of cognitive science and computational optimization. A detailed complexity analysis confirms that HPPSO retains the same time and space complexity as standard PSO. Convergence analysis further demonstrates its ability to prevent premature stagnation and consistently approach near-global optima. Extensive experiments on 20 benchmark functions—including original and shifted variants—under multiple dimensional settings (10, 100, and 2000) show that HPPSO consistently outperforms existing metaheuristic algorithms, particularly in high-dimensional and rugged landscapes. Additionally, HPPSO was applied to neural network training tasks for classification and regression, where it achieved the lowest training error among all compared metaheuristic optimizers. This study contributes a fundamentally new paradigm in nature-inspired optimization by embedding principles of human personality into algorithmic design.