Pengenalan Genus Diatom Menggunakan Principal Component Analysis dan Jaringan Syaraf Tiruan Backpropagation sebagai Classifier

Abstract

Diatoms are unicellular algae which have a size between 2 μm and 4 mm. Their importance resides in the fact that they can be used in several research and scientific fields. For instance, they can be used to measure sediment pH, medicinal, as water quality indicators, etc. The recognition and identification of diatoms is a tedious work. This classification process is complicated even for the experts, because there are hundreds of different taxa with many variations in shapes and biological characteristics. This research apply Principal Component Analysis (PCA) for data reduction and Artificial Neural Network (ANN) to identify some kind of diatoms. The proportion of PCA are used in this research is 80% and 90%. This proportion is considered to replace the original data without much loss of information. Backpropagation ANN that used is a single hidden layer. The data used in this study is a JPG image of diatom sampling using electric microscope taken from Biomikro Laboratory, The productivity and Water Environment, Department of Water Resources Management, Faculty of Fisheries and Marine Science, Bogor Agricultural University. All images are divide using two scenarios percentage. The first scenario divide 60% for training data and 40% for testing data, while the second percentage is 80% for training data and 20% of testing data. As result, ANN can be used to identify diatoms. The results showed that the best generalization rate of 90% was obtained in an experiment using 90% PCA proportion with 80% of training data and 20% of testing data.