A fast graphical similarity algorithm for pattern recognition for data from a voltammetric electronic tongue

Abstract

Multi-sensor arrays have been applied more and more extensively in various fields, especially in the electrochemical industry. Among them, cross-sensitive multi-sensor arrays (e.g. electronic tongues and electronic noses) are the most popular ones. Taking advantage of the cross-sensitive response of voltammetric electronic tongues, a unique fast approach for variable reduction and pattern recognition is proposed in this paper based on a set of well-designed graphical similarity principles. It translates the measurement data of electronic tongues into a net graph containing rich pattern information, and then compares the similarity between the graphs on the levels of topology, geometric shape, and geometric size respectively and finally gives the corresponding similarity distances between the samples for pattern recognition.