Classification of Coffee Roast Levels using Multi-Layer Perceptron on MNDT s-Parameters

• Published in: IEEE Symposium on Wireless Technology and Applications, ISWTA
• Main Author: Salim A.; Yassin I.M.; Mahmood M.K.A.; Khan Z.I.; Ali M.S.A.M.; Mohd Shariff K.K.; Osman F.N.; Ahmad A.; Eskandari F.
• Format: Conference paper
• Language: English
• Published: IEEE Computer Society 2022
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85142730037&doi=10.1109%2fISWTA55313.2022.9942782&partnerID=40&md5=a6143456ae82e693bb2237ce8d0e6c6e
• ISSN: 23247843
• DOI: 10.1109/ISWTA55313.2022.9942782

This paper presents an intelligent method to differentiate levels of roasting of coffee based on data collected from Microwave Non-Destructive Testing (MNDT) method. The MNDT method beams microwaves through several types of coffee (dark, medium, and light roast) and obtains the s-parameter readings from them. A multi-layer perceptron neural network was then then fed to the MLP to determine the degree of different coffee roasts. The MLP could differentiate between the different roasts (with 6,400 data points per sample) with a modest number of hidden units. © 2022 IEEE.