Kohonen neural nets

Self-organizing ANNs (Kohonen neural nets) were employed for classifying different steels [88]. Twelve relevant elements were selected for data processing through ANNs. 

A later chapter will discuss these methods in more detail. For example, support vector machines and traditional neural networks are analogs of multiple regression or discriminant analysis that provide more flexibility in the form of the relationship between molecular properties and bioactivity.Kohonen neural nets are a more flexible analog to principal component analysis. Various Bayesian approaches are alternatives to the statistical methods described earlier. A freely available program oflcrs many of these capabilities. ... 

Some of the pioneering studies published by several reputed authors in the chemometrics field [55] employed Kohonen neural networks to diagnose calibration problems related to the use of AAS spectral lines. As they focused on classifying potential calibration lines, they used Kohonen neural networks to perform a sort of pattern recognition. Often Kohonen nets (which were outlined briefly in Section 5.4.1) are best suited to perform classification tasks, whereas error back-propagation feed-forwards (BPNs) are preferred for calibration purposes [56]. 

Fig. 10.11 Representation of hybrid approach in which Kohonen maps, neural nets, multiple component analysis, and pattern recognition are combined to create a complex data evaluation cascade. Within this cascade supervised (quantitative) and unsupervised (qualitative) techniques are combined (Hierlemann et al., 1996)...

In the case of unsupervised learning, only input vectors are presented as known from factorial methods and cluster analysis (cf. Section 5.2). The objects are grouped on the basis of their features. Unknown objects can then be automatically recognized. We will learn about the Kohonen network as a typical neural net of that kind. 

Neural nets can also be based on unsupervised learning strategies. To date these nets have been employed primarily to support data visualization, but their flexibility is such that they are becoming more common in a wide variety of applications. A simple version of an unsupervised neural net is the Kohonen self-organizing map (SOM) (Kohonen, 1982, 1984 Lang, this volume). These nets also use a set number of nodes, but operate according to different principles. 

IR Recognition of substructures and compound classes Spectra-structure relationships, PCA, PLS, Kohonen map Spectra prediction, neural net Drug identification 93-95 96-98 99, 100 101... 

Kohonen, T. A Simple Paradigm For The Self-Organized Formatiom Of Structured Feature Maps In Competition And Cooperation In Neural Nets (Lecture notes in biomathematics vol 45, Amari, S. Arbib, M. A. Eds.) ISBN 0387115749 Springer-Verlag Berlin, 1982. 

A specialized method for similarity-based visualization of high-dimensional data is formed by self-organizing feature maps (SOM). The data items are arranged on a two-dimensional plane with the aid of neural networks, especially Kohonen nets. Similarity between data items is represented by spacial closeness, while large distances indicate major dissimilarities [968]. At the authors department, a system called MIDAS had already been developed which combines strategies for the creation of feature maps with the supervised generation of fuzzy-terms from the maps [967]. 

Neural networks for unsupervised learning are based on a competitive layer of weights arranged linearly or in a plane (Figure 8.13). If arranged in a plane, the nets are termed a Kohonen network. The peculiarity of this network type is the maintenance of topology or, more general, the pattern of the data vector to be learned. 

---