Classification of Chili Plant Origin by Using Multilayer Perceptron Neural Network

The geographical origin of the plants can affect the growth and hence the quality of the plants. In this research, the origin of the chili plants has been investigated by using Fourier transform infrared (FTIR) spectroscopy. The spectroscopy generated 3734 data with a wavenumber range from 4000–400 cm−1. The pre-processing of the spectra was done by using baseline correction and vector normalization. The analysis was then taken in the biofingerprint area of 1800–900 cm−1 range which has 934 data points. Feature extraction for dimension reduction was achieved using principal component analysis (PCA). The PC scores from PCA were then fed into a k-means and a multilayer perceptron neural network (MLPNN). The k-means clustering shows that the samples can be distinguished into three different groups. Meanwhile, for the MLPNN, the number of the hidden layer's neurons and the learning rate of the system were optimized to get the best classification result. A hidden layer with twenty neurons had the highest accuracy, while a learning rate of 0.001 had the highest value of 100%.