Weighted C-type random 2 satisfiability in discrete hopfield neural network

In order to better solve real-world optimization problems, Boolean Satisfiability logic rules combined with the Discrete Hopfield Neural Network offers effective methods in the area of Artificial Intelligence. This article proposes a flexible logical rule called Weighted C-Type Random 2 Satisfiabili...

Full description

Saved in:
Bibliographic Details
Main Author: Chang, Yunjie
Format: Conference Proceeding
Language:English
Subjects:
Artificial intelligence
Error correction
Neural networks
Performance evaluation
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In order to better solve real-world optimization problems, Boolean Satisfiability logic rules combined with the Discrete Hopfield Neural Network offers effective methods in the area of Artificial Intelligence. This article proposes a flexible logical rule called Weighted C-Type Random 2 Satisfiability that is embedded into Discrete Hopfield Neural Network. In Weighted C-Type Random 2 Satisfiability, the initial ratio of negative literal is pre-determined, and the first-order clauses, the second-order clauses, or both are randomly generated. In order to assess the performance of proposed model, several experiments were conducted through simulation environment using various performance evaluation metrics, such as the average iteration time, the ratio of correct logical rules, mean absolute error and Similarity index. Experimental results showed that Weighted C-Type Random 2 Satisfiability has advantages in terms of learning property, convergence property and solution diversity property, and it can be applied in various real-life fields that require random dynamics analysis.
ISSN:0094-243X
1551-7616
DOI:10.1063/5.0224838