The effect of logical permutation in 2 satisfiability reverse analysis method

Abstract

The dynamical behaviors of logic mining in real datasets are strongly dependent by its logical structure. In this case, logical rule that has been embedded to neural network has long suffered from a lack of interpretability and accuracy. This has severely limited the practical usability of logic mining. Logical permutation is a definitive finite arrangement of attributes that makes 2SAT became true. It was believed that the effect of permutation will increase the accuracy of the system. In this paper, we presented the effect of logical permutation in logic mining (2SATRA) integrated with recurrent Hopfield Neural Network (HNN). Several benchmark datasets will be used to validate the effect of logical permutation. It has been shown that 2SATRA with different permutation will results in improvement in terms of accuracy value. This finding will lead to a better understand of 2SATRA in doing real life datasets.