S-Type Satisfiability Logic Mining for Medical Datasets
Suad Abdeen, Mohd Shareduwan Mohd Kasihmuddin, Mohd. Asyraf Mansor, Nur Ezlin Zamri, Nurul Atiqah Romli · 2025
The logic mining approach has been explored extensively by various researchers. Nevertheless, current logic mining algorithms have neglected the significance of data preprocessing, resulting in a limited capacity to generalize the retrieved induced logic. Furthermore, present logic mining models have significant disadvantages, such as rigid logical structures. This work addresses the existing gap by proposing a new logic mining model. The model combines a supervised data preprocessing phase with the non-systematic satisfiability of S-type Random 2 Satisfiability within a discrete Hopfield neural network. The innovative dynamic-unit discrete Hopfield neural network integrates a multi-objective function, which greatly improves the search space and results in optimal solutions. The medical dataset experiments and performance metrics indicate that the proposed model surpasses its counterparts.