Identification of agarwood oil qualities using radial basis function (RBF) neural network / N. S. A. Zubir ...[et al.]

• Main Authors: Zubir, N. S. A., Abas, M. A., Ismail, Nurlaila, M.Ali, Nor Azah, Rahiman, M. H. F., Ng, K. M., Saiful, N. T., Taib, M. N.
• Format: Article
• Language: English
• Published: UiTM Press 2017
• Subjects: Pattern recognition systems
• Online Access: https://ir.uitm.edu.my/id/eprint/63010/1/63010.pdf; https://ir.uitm.edu.my/id/eprint/63010/; https://jeesr.uitm.edu.my/v1/
• Institution: Universiti Teknologi Mara

Agarwood is known as a valuable non-timber product found in the dark fragrant resin in the stem, branch and roots of certain species of Aquilaria. Agarwood oil is one of the popular essential oil that has been used not only in Asian but in the world. The price of the agarwood oil is referring based on the quality of agarwood oil. The agarwood oil have distinct pattern which can be discriminating the qualities of agarwood oil by classification technique such as radial basis function. The Radial Basis Function networks (RBFNs) are commonly used for complex pattern classification. This study examines the performance of radial basis function of identifying the quality of agarwood oil either high or low quality. The dataset consists of the abundances of significant compounds (%) and qualities of the agarwood oil. The result reveals that the classification using RBF technique, performs slightly have a better performance of MSE values depends on the 100 maximum numbers of neurons and 3 number of spread. The hypothesis from this study is the larger number of spread the smoother the function approximation. Besides that, the small number of spread the large number of neurons required to fit a smooth function.