The paper proposes a new feature selection strategy based on rough sets and Particle Swarm Optimization (PSO). It addresses the limitations of hill-climbing approaches in rough set theory, which often fail to find optimal reductions due to the lack of a perfect heuristic. In contrast, PSO, an evolutionary computation technique, is shown to be effective in finding optimal feature subsets or reducts. The algorithm is evaluated using UCI datasets and compared with other algorithms, including genetic algorithms (GA) and deterministic rough set reduction algorithms. The results demonstrate that PSO is efficient and robust, capable of discovering optimal solutions quickly. The paper also discusses the implementation details of the PSO algorithm for feature selection, including the representation of positions and velocities, position update strategies, and the fitness function. Experimental results show that PSO outperforms GA in terms of both solution quality and computational efficiency. The study concludes that PSO is a promising method for rough set reduction, with potential for further improvement and application in various fields.The paper proposes a new feature selection strategy based on rough sets and Particle Swarm Optimization (PSO). It addresses the limitations of hill-climbing approaches in rough set theory, which often fail to find optimal reductions due to the lack of a perfect heuristic. In contrast, PSO, an evolutionary computation technique, is shown to be effective in finding optimal feature subsets or reducts. The algorithm is evaluated using UCI datasets and compared with other algorithms, including genetic algorithms (GA) and deterministic rough set reduction algorithms. The results demonstrate that PSO is efficient and robust, capable of discovering optimal solutions quickly. The paper also discusses the implementation details of the PSO algorithm for feature selection, including the representation of positions and velocities, position update strategies, and the fitness function. Experimental results show that PSO outperforms GA in terms of both solution quality and computational efficiency. The study concludes that PSO is a promising method for rough set reduction, with potential for further improvement and application in various fields.