Analysis of Horn 3-satisfiability logical structure in Hopfield neural network
Gaeithry Manoharam, Mohd Shareduwan Mohd Kasihmuddin, Mohd. Asyraf Mansor, Siti Zulaikha Mohd Jamaludin, Nur Ezlin Zamri · AIP conference proceedings · 2024
The Propositional Satisfiability (SAT) is an optimization problem that give solution in many fields such as planning, verification, and security.The SAT solver and Neural Network are the two greatest achievement in computer science to made significant contributions in the variety of real-world issues.In the Hopfield Neural Network (HNN), nonsystematic and systematic SAT used as a theoretical input.This study introduces the systematic logical structure Horn 3-SAT (Horn 3-SAT) in HNN by simply add the number of iterations.The Horn 3-SAT gives solutions in larger and more difficult in the training phase.The Horn3SAT formulas are Boolean expressions with the feature that each sentence can only include one positive variable.They are expressed in conjunctive normal form.The logical framework when all clauses are met, Horn 3-SAT calculates the minimal cost function and saves the truth values of the atoms in neurons.Horn 3-SAT's functionality and scalability in HNN are evaluated using performance metrics.A novel logical system for data mining that combines a dynamic of literals and clauses is often proposed by the Horn 3-SAT formulation.The global minimum of the energy function indicates stable configuration.Modification in the logical structure can increase the quality of the solution by escape from local minima in the searching process of the global optimum.