Major 2 Satisfiability Logic in Discrete Hopfield Neural Network. Issue 5 (4th May 2022)

• Record Type: Journal Article
• Authors: Alway, Alyaa; Zamri, Nur Ezlin; Karim, Syed Anayet; Mansor, Mohd. Asyraf; Mohd Kasihmuddin, Mohd Shareduwan; Mohammed Bazuhair, Muna
• Is Part Of: International journal of computer mathematics. Volume 99:Issue 5(2022)
• Page Start: 924
• Page End: 948
• Publication Date: 2022-05-04
• Subjects: Discrete Hopfield neural network -- Major 2 Satisfiability -- artificial neural network -- satisfiability logic -- symbolic integration; 68T07 -- 68R07; Computers -- Periodicals; Numerical analysis -- Periodicals; Automation -- Periodicals; 004.0151
• DOI: 10.1080/00207160.2021.1939870
• Languages: English
• ISSNs: 0020-7160

Abstract

Abstract : Existing satisfiability (SAT) is composed of a systematic logical structure with definite literals in a set of clauses. The key problem of the existing SAT is the lack of interpretability of a logical structure that leads to low variability of the retrieved neuron states. Thus, a new non-systematic SAT with higher interpretability is needed to reduce the repetition of the patterns of the final neuron states. This paper presents Major 2 Satisfiability (MAJ2SAT) by emphasizing a ratio of 2 Satisfiability (2SAT) clauses present in non-systematic SAT. Hence, different compositions of MAJ2SAT are implemented in a Discrete Hopfield Neural Network (DHNN) by adopting an Exhaustive Search as a training algorithm. Various performance metrics are utilized to measure the compatibility and behaviour of MAJ2SAT in DHNN. Overall, the formulation of MAJ2SAT offers an alternative logical structure in the field of data mining that involves a more dynamic composition of literals and clauses.