Identification of Salt Production using Multi-Layer Perceptron and Convolutional Neural Network

  • N. Bala Vignesh, Dr. V. Joseph Peter
Keywords: Salt Production identification, Convolutional Neural Network (CNN), Multi-Layer Perceptron (MLP), Deep Learning, Artificial Intelligence.

Abstract

Accurate and automated identification of production from salt informatics is essential to improve manufacturing in salt industries. In this research, two different classification methods like Multi-Layer Perceptron (MLP) and Convolutional Neural Network (CNN) are processed over the salt dataset to extract production features automatically. The outcome features from MLP and CNN have been fed separately into softmax classification for predicting production class labels. These labels are compared with original class labels from the salt dataset to perform an evaluation. Thus, the MLP method achieves 83% of accuracy, 83% of F1-Score, 85% of Recall, 87% of precision and the CNN method achieves 86% of accuracy, 91% of F1-Score, 91% of Recall and 91% of precision values. From the experimental comparison, the CNN architecture achieves 3.6% higher accuracy, 9.6% higher F1-score, 7.1% higher recall and 4.6% higher precision values when compared to the MLP architecture.

Published
2021-08-07
How to Cite
Dr. V. Joseph Peter, N. B. V. (2021). Identification of Salt Production using Multi-Layer Perceptron and Convolutional Neural Network. Design Engineering, 6906- 6912. Retrieved from http://www.thedesignengineering.com/index.php/DE/article/view/3207
Section
Articles