Pattern recognition algorithms

In the field of chemical sensors, the revolution in software and inexpensive hardware means that not only nonlinear chemical responses can be tolerated, but incomplete selectivity to a variety of chemical species can also be handled. Arrays of imperfectly selective sensors can be used in conjunction with pattern recognition algorithms to sort out classes of chemical compounds and thek concentrations when the latter are mixed together. 

The successful appHcation of pattern recognition methods depends on a number of assumptions (14). Obviously, there must be multiple samples from a system with multiple measurements consistendy made on each sample. For many techniques the system should be overdeterrnined the ratio of number of samples to number of measurements should be at least three. These techniques assume that the nearness of points in hyperspace faithfully redects the similarity of the properties of the samples. The data should be arranged in a data matrix with one row per sample, and the entries of each row should be the measurements made on the sample, as shown in Figure 1. The information needed to answer the questions must be implicitly contained in that data matrix, and the data representation must be conformable with the pattern recognition algorithms used. 

The multivariate tools typically used for the NIR-CI analysis of pharmaceutical products fall into two main categories pattern recognition techniques and factor-based chemometric analysis methods. Pattern recognition algorithms such as spectral correlation or Euclidian distance calculations basically determine the similarity of a sample spectrum to a reference spectrum. These tools are especially useful for images where the individual pixels yield relatively unmixed spectra. These techniques can be used to quickly define spatial distributions of known materials based on external reference spectra. Alternatively, they can be used with internal references, to locate and classify regions with similar spectral response. 

Another example is LIBS application for real-time identification of carious teeth (Samek et al. 2003). In the dental practice, usually more healthy tissue is removed than ultimately necessary. Carious and healthy tooth material can be identified through the decrease of matrix elements Ca and P in hydroxyapatite and/or the increase of non-matrix elements, typically Li, Sr, Ba, Na, Mg, Zn and C, using pattern recognition algorithms. A fiber-based LIBS assembly was successfully used for this task. As for the case of phosphate ores evaluation, the efforts aimed at normalizing the spectrum collection conditions and procedures, so that the spectra are sufficiently reproducible for precise quantitative... 

A generalised structure of an electronic nose is shown in Fig. 15.9. The sensor array may be QMB, conducting polymer, MOS or MS-based sensors. The data generated by each sensor are processed by a pattern-recognition algorithm and the results are then analysed. The ability to characterise complex mixtures without the need to identify and quantify individual components is one of the main advantages of such an approach. The pattern-recognition methods maybe divided into non-supervised (e.g. principal component analysis, PCA) and supervised (artificial neural network, ANN) methods also a combination of both can be used. 

Graphical methods in connection with pattern recognition algorithms, i.e. geometrical or statistical methods, e.g. minimum spanning tree or cluster analysis, are more powerful methods for explorative data analysis than graphical methods alone. 

Fig. 2. The principle configuration of an electronic nose system where the analyte mixture is contacted with a chemical sensor array that produces raw data which subsequently are treated with a pattern recognition algorithm that delivers the predicted result...

As most pattern recognition algorithms use conventional geometric parameters of ellipses, namely semi-major and semi-minor axes and eccentricity, this section will deduce all the necessary transformations between the... 

---