Metal Oxide Semiconductor Gas Sensors-based E-nose and Two-stage Classification: Authentication of Malaysia and Vietnam Black Pepper Samples

The superior quality of black pepper from Sarawak, Malaysia highlights the importance of geo-tracing as a mean to establish product differentiation. The geo-tracing method developed should not only differentiate between Malaysia and non-Malaysia black pepper, but also specify the country of origin,...

Full description

Saved in:
Bibliographic Details
Published in2022 IEEE International Symposium on Olfaction and Electronic Nose (ISOEN) pp. 1 - 4
Main Authors En Lee, Hui, Mercer, Zehnder J. A., Ng, Sing Muk, Shafiei, Mahnaz, Chua, Hong Siang
Format Conference Proceeding
LanguageEnglish
Published IEEE 29.05.2022
Subjects
Analytical models
black pepper
classification
electronic nose (e-nose)
Electronic noses
gas sensors
machine learning
metal oxide semiconductor (MOS)
Metals
Predictive models
Semiconductor device modeling
Support vector machines
Training
Online AccessGet full text

Cover

More Information
Summary:The superior quality of black pepper from Sarawak, Malaysia highlights the importance of geo-tracing as a mean to establish product differentiation. The geo-tracing method developed should not only differentiate between Malaysia and non-Malaysia black pepper, but also specify the country of origin, which is the deciding factor of price. This study has developed a two-stage classification model trained from responses of an electronic nose (e-nose) comprising of four metal oxide semiconductor (MOS) gas sensors and black pepper mass to authenticate Malaysia and Vietnam black pepper from India, Indonesia and frauded samples. 24 classifiers were trained and compared in terms of classification accuracy. By trial and error, 90 s mark of the sampling process has been determined to be the earliest sensors response time that contributes to 100 % accuracy in classifying training dataset and predicting test samples that have been prepared to verify the classification model. In stage one, fine gaussian support vector machine (SVM), weighted k-nearest neighbor (KNN), bagged trees, subspace KNN and random under-sampling (RUS) boosted trees are the classifiers that have authenticated Malaysia samples with 100% accuracy. In stage two, bagged trees classifier has authenticated Vietnam samples from the remaining non-Malaysia samples with 100% accuracy.
DOI:10.1109/ISOEN54820.2022.9789618