Computational Intelligence Paradigms: Innovative Applications (Studies in Computational Intelligence, 137)

System designers are faced with a large set of data which has to be analysed and processed efficiently. Advanced computational intelligence paradigms present tremendous advantages by offering capabilities such as learning, generalisation and robustness. These capabilities help in designing complex systems which are intelligent and robust. The book includes a sample of research on the innovative applications of advanced computational intelligence paradigms. The characteristics of computational intelligence paradigms such as learning, generalization based on learned knowledge, knowledge extraction from imprecise and incomplete data are the extremely important for the implementation of intelligent machines. The chapters include architectures of computational intelligence paradigms, knowledge discovery, pattern classification, clusters, support vector machines and gene linkage analysis. We believe that the research on computational intelligence will simulate great interest among designers and researchers of complex systems. It is important to use the fusion of various constituents of computational intelligence to offset the demerits of one paradigm by the merits of another. Title Page Preface Contents An Introduction to Computational Intelligence Paradigms Introduction Computational Intelligence Models Neural Computational Models Fuzzy Computational Models Evolutionary Computational Models Hybrid Computational Models Chapters Included in This Book Summary Resources Journals Special Issue of Journals Conferences Conference Proceedings Book Series Books Book Chapters References A Quest for Adaptable and Interpretable Architectures of Computational Intelligence Introductory Comments Generic Logic Processing Realized with the Aid of Logic Neurons Triangular Norms: t- Norms Triangular Norms: t-Conorms Uninorms – A Hybrid Structure of Logic Operators Logic Neurons Aggregative Neurons: OR and AND Neurons Referential (Reference) Neurons Uninorm-Based Logic Neurons Architectures of Logic Networks Logic Processor in the Processing of Fuzzy Logic Functions: A Canonical Realization Heterogeneous Logic Networks Unineuron-Based Topologies of Logic Networks Learning of Unineurons and Networks of Unineurons Linguistic Modeling of Relationships between Information Granules: The Paradigm of Granular (Fuzzy) Models Conclusions References Membership Map: A Data Transformation for Knowledge Discovery Based on Granulation and Fuzzy Membership Aggregation Introduction Background Related Work Data Preprocessing Data Partitioning and Labeling MembershipMap Generation Properties of the Crisp, Fuzzy, and Possibilistic Maps The Crisp MembershipMap The Fuzzy MembershipMap The Possibilistic MembershipMap Exploring the Membership Maps Identifying Seed Points Identifying Noise Points and Outliers Identifying Boundary Points Illustrative Example Discussion Data Labeling for Semi-Supervised Learning Experimental Results Data Sets Membership Generation Identifying Regions of Interest Clustering the Membership Maps Classification Using the Membership Maps Application: Color Image Segmentation Conclusions References Advanced Developments and Applications of the Fuzzy ARTMAP Neural Network in Pattern Classification Introduction Fuzzy ARTMAP Principles and Dynamics Advanced FAM-Based Developments Modifications to FAM FAM-Based New Algorithms Advanced FAM-Based Applications Experimental Evaluation of FAM-Based Algorithms Discussion References Large Margin Methods for Structured Output Prediction Introduction Discriminative Models for Structured Output Learning Structured Output Problems Learning with Structured Outputs Sequence Labeling: From HMMs to Chain CRFs Large Margin Approaches for Structured Output Learning Multiclass SVMs Maximizing the Margin in Structured Output Marginal Variables Method Iterative Method Min-Max Method Generalization Bounds Experimental Results Sequence Labeling Learning Sequence Alignment Learning Sequence Parse Learning Discussion and Conclusions References Ensemble MLP Classifier Design Introduction MLP Classifiers and Ensembles Diversity/Accuracy and MCS Error Correcting Output Coding (ECOC) and Multi-class Problems Examples Benchmark Data Face Data Discussion Conclusion References Functional Principal Points and Functional Cluster Analysis Introduction Preliminaries Definition of Principal Points Definition of Random Functions Definition of Functional Principal Points of Random Functions $K$-Means Functional Clustering Orthonormal Basis Transformation of Functions Some Examples of Functional Principal Points The Case That 2-Dimensional Coefficient Vector of Linear Random Function Following Bivariate Normal Distribution The Case That p-Dimensional Coefficient Vector of Polynomial Random Function Following p-Variate Normal Distribution The Case That Fourier Polynomial Random Function Following Multivariate Normal Distribution Optimal Functional Clustering and Functional Principal Points The Numbers of Local Solutions in Functional k-Means Clustering Summary References Clustering with Size Constraints Introduction The FCM Algorithm Equi-sized Clusters Examples for Uniform Clustering Limiting the Size of Single Clusters Only A Distance Measure for Sets of Prototypes Conclusions References Cluster Validating Techniques in the Presence of Duplicates Introduction Outline of Our Approach Clustering Algorithms Partitioning Around Medoids (PAM) Validation Techniques Silhouette Index Calinski and Harabasz Index Baker and Hubert Index Experimental Evaluation Coefficient Relibility wrt Record Based Duplicates Coefficient Reliability wrt Value Based Duplicates Conclusion Fuzzy Blocking Regression Models Introduction Fuzzy Clustering Variable Based Fuzzy Clustering Variable Based Dissimilarity Variable Based Fuzzy Clustering Methods Blocking Regression Model Fuzzy Blocking Regression Model Fuzzy Blocking Regression Model Using Fuzzy Intercepts Variable Based Fuzzy Blocking Regression Model Numerical Examples Conclusion References Support Vector Machines and Features for Environment Perception in Mobile Robotics Introduction Support Vector Machines VC Dimension Multiclassification with SVM Practical Considerations Features Features in Lidar Space Features in Camera Space: A Brief Survey Applications Learning to Label Environment Places Recognizing Objects in Images Conclusion References Linkage Analysis in Genetic Algorithms Introduction Linkage Identification in Genetic Algorithms Problem Decomposability Linkage Identification by Perturbation D$^5$: Dependency Detection by Fitness Clustering Context Dependent Crossover Conclusion References Author Index