Pattern Recognition

From MaRDI portal





Pattern recognition and machine learning form a major area of research and development that encompasses the processing of pictorial, and other non-numerical information obtained from interaction between science, technology and society. A motivation for this spur of activity in this field is the need for the people to communicate with computing machines in their natural mode of communication. Another important motivation is that scientists are also concerned with the idea of designing and making intelligent machines that can carry out certain tasks as we human beings do. The present volume provides a collection of twenty one articles including an introductory chapter, containing both review and new material, describing in a unified way, the recent development of various methodologies, ranging from classical to modern, with significant real life applications. These articles, written by different experts over the world, demonstrate the significance of this evolution relevance of different approaches with characteristic features and the formulation of various modern theorem. The volume starts with an introductory article written by the editors themselves, explaining the basic concepts of pattern recognition, different tasks involved, some conventional classification techniques and the subsequent development of modern methodologies. The significance of data mining, which has recently drawn the attention of many researchers all over the world, is also mentioned. This chapter is included for the convenience of understanding the theory of pattern recognition and hence the remaining chapters of the book. Chapter 2-5 deal with decision theoretic classification using statistical approach. T. Krishnan considers in Chapter 2 the problem of supervised learning based on the multivariate normal model with a common covariance matrix leading to Fisher's linear discriminant function. The effect of imperfect supervision is studied and the efficiency of learning under two imperfect situations, namely, deterministic but error-prone, and stochastic, is determined as compared to perfect supervision. Chapter 3 of M. A. L. Thathachar and P. S. Sastry addresses some adaptive stochastic algorithms for finding optimal decision rules based on learning automaton models. Chapter 4 deals with unsupervised learning of classifiers. Here A. Pal, one of the editors, describes some significant recent Bayesian approaches that can be applied to the unsupervised classification problem, if it is formulated as one of decomposition of finite mixtures of probability densities. In Chapter 5, K. V. Mardia focuses on a specific problem of pattern recognition, namely, recognition of shapes in an image. The author presents an excellent review of the recent advances with emphasis on high level Bayesian image analysis. In Chapter 6, R. Kothari and M. Dong provide an overview of decision tree induction algorithms and approaches. The limitation of the decision approach lies in its incapability to articulate the interrelationship of the pattern substructures. This has lead to the development of syntactic or structural pattern recognition. Chapter 7 by A. K. Majumdar and A. K. Roy provides a fairly extensive review on such an approach considering its different facets, e.g. primitive selection, pattern grammars, etc. An application to character recognition is also shown. When the pattern indeterminacy is due to inherent vagueness rather than randomness, fuzzy set theoretic approach becomes useful. Chapter 8 of W. Pedrycz and N. Pizzi presents a motivation for the use of fuzzy sets as a logic canvas for pattern recognition problems with imprecise information. The authors focus on fuzzy set based generalisation of logic processors and Petri nets as well as on the contex-based partially supervised extensions to the fuzzy c-means clustering. In Chapter 9, M. Grabisch describes the role of fuzzy integrals and fuzzy measures in the development of two new methodologies for pattern recognition. Artifical neural network based methods have characteristics like adaptivity, speed, robustness and optimality. Chapters 10-12 deal with pattern recognition in neural computing paradigm. V. David Sanchez A. describes in Chapter 10, the relationship between traditional methods and neural networks for pattern recognition. Chapter 11 of N. B. Karayiannis and his colleagues R. Kretzschmar and H. Richner focuses on the development, testing and evaluation of pattern classifiers using quantum neural networks. Chapter 12 concentrates on the application of networks of spiking neurons in data mining problems. Here the authors, K. Cios and D. M. Sala, present some new findings in design and applications of neural networks that use a biologically inspired spiking neuron model. Chapter 13 shows such an attempt in designing a nonparametric classifier. Chapter 14 is based on rough set theory which has emerged as another mathematical approach for dealing with imprecise or vague concepts arising mainly from the granularity in the universe of discourse. The next three chapters deal with synergistic integration of the merits of fuzzy logic, neural networks and genetic algorithms in order to build more efficient recognition system under soft computing framework. Chapter 15 of L. I. Kuncheva explains how the aforesaid three technologies can be used in classifier combination, which is an established subdiscipline of pattern recognition. The problem of presenting complex objects and retrieving ``interesting structures from a large database has been addressed in Chapter 16 by E. Ruspini and I. J. Zwir in the framework of soft computing. Chapter 17 concerns about the neuro-fuzzy approach which is considered to be the most visible hybridization, so far made, in soft computing paradigm. Chapter 18-21 demonstrate various applications. Chapter 18 deals with adaptive segmentation techniques for hyperspectral imagery. Chapter 19 is concerned with pattern recognition issues in speech processing. The problem of on-line handwriting recognition is considered in Chapter 20. The last chapter addresses an important problem, namely, tongue diagnosis based on biometric pattern recognition.




Cited in
(27)








This page was built for publication: Pattern Recognition

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3146957)