Data mining and knowledge discovery via logic-based methods: theory, algorithms, and applications

There are many approaches to data mining and knowledge discovery (DM&KD), including neural networks, closest neighbor methods, and various statistical methods. This monograph, however, focuses on the development and use of a novel approach, based on mathematical logic, that the author and his research associates have worked on over the last 20 years. The methods presented in the book deal with key DM&KD issues in an intuitive manner and in a natural sequence. Compared to other DM&KD methods, those based on mathematical logic offer a directand often intuitive approach for extracting easily interpretable patterns from databases. The book discusses the theoretical foundations of the methods described, and it also presents a wide collection of examples, many of which comefrom real-life applications. Almost all theoretical developments are accompanied by extensive empirical analysis which often involved the solution of a very large number of simulated test problems. Using a novel method, the monographstudies a series of interconnected key data mining and knowledge discovery problems Provides a unique perspective into the essence of some fundamental DataMining issues, many of which come from important real life applications Applications and algorithms are accompanied by experimental results INDICE: Foreword.- Preface.- Acknowledgments.- List of Figures.- List of Tables.- Part I. Algorithmic Issues.- 1.Introduction.- 2.Inferring a Boolean Function from Positive and Negative Examples.- 3.A Revised Branch-and-Bound Approach for Inferring a Boolean Function from Examples.- 4.Some Fast Heuristics for Inferring a Boolean Function from Examples.- 5.An Approach to Guided Learning of Boolean Functions.- 6.An Incremental Algorithm for Inferring Boolean Functions.- 7.A Duality Relationship Between Boolean Functions in CNF and DNF Derivable from the Same Training Examples.- 8.The Rejectability Graph of Two Setsof Examples.- Part II. Application Issues.- 9.The Reliability Issue in Data Mining: The Case of Computer-Aided Breast Cancer Diagnosis.- 10.Data Mining andKnowledge Discovery by Means of Monotone Boolean Functions.- 11.Some Application Issues of Monotone Boolean Functions.- 12.Mining of Association Rules.- 13.Data Mining of Text Documents.- 14.First Case Study: Predicting Muscle Fatigue from EMG Signals.- 15.Second Case Study: Inference of Diagnostic Rules for Breast Cancer.- 16.A Fuzzy Logic Approach to Attribute Formalization: Analysis of Lobulation for Breast Cancer Diagnosis.- 17.Conclusions.- References.- Subject Index.- Author Index.- About the Author.

• ISBN: 978-1-4419-1629-7
• Editorial: Springer
• Encuadernacion: Cartoné
• Páginas: 350
• Fecha Publicación: 17/06/2010
• Nº Volúmenes: 1
• Idioma: Inglés