Classification and Mapping

The second step of the algorithm is the comparison of descriptor values. Several different methods are applicable to the problem, most of which are adopted from pattern recognition or statistical sciences. 

For a manual interpretation of the diversity behavior of a dataset, the descriptors 

Purely numerical methods such as hierarchical clustering or the minimal spanning tree compute the similarity of molecules directly in the high-dimensional descriptor space. The results promise higher accuracy (no mapping errors), but their interpretation is less intuitive. 

The chemist can select compounds for synthesis from the maps or clustering results with the possibility of additional optimization such as use of preferred building blocks or functional groups in the molecules. 

In general, it is not necessary for asdentist in drug design to understand the theory of classification methods in detail. The dedsion as to which method is to be used depends mainly on technical aspects, such as size or dimension of the dataset or complexity of the problem. The classification results do not depend heavily on the method used, but especially cluster borders will differ sHghdy. Classification software will work out and visualize the information already included in the descriptors. 

---

Hartman, P. E., Hartman, Z., Stahl, R. C., and Ames, B. N. (1971a), Classification and mapping of spontaneous and induced mutations in the histidine operon of Salmonella, Advan. Genet. 16, 1-34. 

Bucknum [1] in work first described in 1997, outlined a general scheme for the systematic classification and mapping of the polyhedra, 2-dimen-sional tessellations and 3-dimensional networics in a self-consistent topological space for these structures. This general scheme begins with a consideration of the Euler relation [2] for the polyhedra, shown as Eq. (1), which was first proposed in 1758 to the Russian Academy by Euler, and was, in fact, the point of departure for Euler into a new area of mathematics thereafter known explicitly as topology. 

In practice image quality is also reduced by use of high mass resolution and energy offset. Often, therefore, mass interference cannot he avoided. Determination of element distributions is possible by use of image processing tools for classification of mappings of different masses [3.53]. 

Avdeef, A., Kansy, M., Bendels, S., Tsinman, K. Biopharmaceutics classification gradient maps and the pH partition antithesis. Review by Eur. J. Pharm. Sci. 

Villars, P., Mathis, K. and Hulliger, F. (1989) Environment classification and structural stability maps. In The Structures of Binary Compounds, eds. de Boer, F. and Pettifor, D. (North-Holland, Amsterdam), Vol. 2, p. 1. 

The search for regularities and criteria for the synthesis of new representatives of particular structure types has been carried out by many authors. Several factors have been recognized to be important in controlling the structural stability, and some of these were used as coordinates for the preparation of classification and prediction maps in which various compounds can be plotted and separated into different structure domains. 

Bendels, S., Tsinman, O., Wagner, B., Lipp, D., Parrilla, 1., Kansy M. and Avdeef, A. (2006) PAMPA-excipient classification gradient map. Pharmaceutical Research, 23, 2525-2535. 

Support Vector Machine (SVM) is a classification and regression method developed by Vapnik.30 In support vector regression (SVR), the input variables are first mapped into a higher dimensional feature space by the use of a kernel function, and then a linear model is constructed in this feature space. The kernel functions often used in SVM include linear, polynomial, radial basis function (RBF), and sigmoid function. The generalization performance of SVM depends on the selection of several internal parameters of the algorithm (C and e), the type of kernel, and the parameters of the kernel.31... 

The next step was to use a Kohonen two-dimensional self-organizing map to represent the spectral data in the 218-dimensional measurement space. Self-organizing maps are, for the most part, used to visualize high-dimensional data. However, classification and prediction of multivariate data can also be performed with these... 

Self-organizing maps in conjunction with principal component analysis constitute a powerful approach for display and classification of multivariate data. However, this does not mean that feature selection should not be used to strengthen the classification of the data. Deletion of irrelevant features can improve the reliability of the classification because noisy variables increase the chances of false classification and decrease classification success rates on new data. Furthermore, feature selection can lead to an understanding of the essential features that play an important role in governing the behavior of the system or process under investigation. It can identify those measurements that are informative and those measurements that are uninformative. However, any approach used for feature selection should take into account the existence of redundancies in the data and be multivariate in nature to ensure identification of all relevant features. 

We can sum up what one can do with a neural network. In principle, neural networks are universal approximators and can compute any computable function. In practice, neural networks are especially useful for classification and function approximation/mapping problems that have plenty of training data available and can tolerate some imprecision but that resist the easy application of hard and fast rules. 

Kohonen self-organizing map An unsupervised learning method of clustering, based on the k-means algorithm, similar to the first stage of radial basis function networks. Self-organized maps are used for classification and clustering. 

Soils were mapped in the Brazilian classification and only the higher taxa can be correlated accurately with the U.S. Soii Taxonomy (Beinroth 1975). 

During soil survey operations, numerous soil profile descriptions, and large amounts of laboratory data, are generated for the soils that are being mapped. Generally, the soils being described, sampled, and analyzed are representatives of soil series, the most detailed (and restrictive) classification of soils in the USA. There are —2.1 X 10" soU series that have been identified and mapped in the USA. The locations, soil descriptions, and... 

Cave sediments were differentiated based on color, grain size, and general composition. Sediments were mapped and digitally drawn onto the plan and profile cave maps. The classification and descriptions of mapped sediments are shown in Table 1. The distribution of sediments were mapped onto cave profiles Ato determine the vertical... 

---