Mitigating Saturation in Fuzzy-Flip-Flop Neural Networks Trained with Memetic PSO Algorithm

This paper investigates the efficacy of learning algorithms for training Fuzzy Flip-Flop Neural Networks, focusing on Particle Swarm Optimization (PSO) and memetic PSO variants. The study explores their performance across diverse datasets, including Breast Cancer Wisconsin, Seeds, and Glass. Results demonstrate that memetic PSO variants, consistently outperform other algorithms in terms of classification accuracy and reduced saturation levels. The examination of different crossover strategies within the memetic PSO framework provides insights into their impact on learning performance. Overall, this research highlights the superiority of memetic PSO algorithms in optimising Fuzzy Flip-Flop Neural Networks for classification tasks with complex datasets, suggesting their potential for broader applications.