期刊名称：International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
印刷版ISSN：2278-1323
出版年度：2013
卷号：2
期号：5
页码：1684-1688
出版社：Shri Pannalal Research Institute of Technolgy
摘要：Clustering is an important task for machine learning which gives best discriminability among different subsets of features. It is usually a Classification problem with unsupervised learning paradigm. Recently unsupervised learning paradigms have gained tremendous attention, especially in the field of electrochemistry, bioinformatics. A novel impedance Tongue (i-Tongue) employing non specific multi- electrode electrochemical impedance spectroscopy is used for classification of Indian black tea. Impedance response at logarithmic frequency interval (features) ranging from 15 MHz (high frequency range) to 20 Hz (low frequency range) of three different type of electrodes were measured by using standard electrochemical workstation, which is used as our features dataset. Further the dimensions of these feature dataset containing impedances at particular frequency intervals are reduced by using Principal Component Analysis (PCA). Our proposed algorithm uses features similarity to distinguish between different tea samples by using a K-Means Clustering as a classifier to find the optimal data locations to have the best discriminability with minimum intra-cluster distance and maximum inter-cluster distance among different tea classes.
