Enhancing Discrete Hopfield Neural Network with Exempted Random 3-Satisfiability and Metaheuristic Learning

The results highlight how the newly proposed Exempted Random 3 Satisfiability (eRAN3SAT) logic performs when integrated into the Discrete Hopfield Neural Network framework (DHNN-eRAN3SAT). This model is further strengthened by three metaheuristic learning techniques: the Election Algorithm, Exhaustive Search, and the Genetic Algorithm. To understand the model’s overall behaviour, we evaluate it using several important metrics, including RMSE learning, RMSE testing, Global Minima Ratio, Total Variation (TV), and the Jaccard Similarity Index (JSI). These results are presented for three clause combinations: k = 1, 2, 3; k = 1, 3; and k = 2, 3, enabling a more precise comparison of the model’s performance across different logical structures.