AsiaSim 2014: 14th International Conference on Systems Simulation, Kitakyushu, Japan, October 26-30, 2014. Proceedings
Design of water control structures, reservoir management, economic evaluation of flood protection projects, land use planning and management, flood insurance assessment, and other projects rely on knowledge of magnitude and frequency of floods. Often, estimation of floods is not easy because of lack of flood records at the target sites. Regional flood frequency analysis (RFFA) alleviates this problem by utilizing flood records pooled from other watersheds, which are similar to the watershed of the target site in flood characteristics.
Clustering techniques are used to identify group(s) of watersheds which have similar flood characteristics. This book is a comprehensive reference on how to use these techniques for RFFA and is the first of its kind. It provides a detailed account of several recently developed clustering techniques, including those based on fuzzy set theory and artificial neural networks. It also documents research findings on application of clustering techniques to RFFA that remain scattered in various hydrology and water resources journals.
The optimal number of groups defined in an area is based on cluster validation measures and L-moment based homogeneity tests. These form the bases to check the regions for homogeneity.
The subjectivity involved and the effort needed to identify homogeneous groups of watersheds with conventional approaches are greatly reduced by using efficient clustering techniques discussed in this book. Furthermore, better flood estimates with smaller confidence intervals are obtained by analysis of data from homogeneous watersheds. Consequently, the problem of over- or under-designing by using these flood estimates is reduced. This leads to optimal economic design of structures. The advantages of better regionalization of watersheds and their utility are entering into hydrologic practice.
This book will be of interest to researchers in stochastic hydrology, practitioners in hydrology and graduate students.
The first five chapters of this volume investigate advances in the use of instance-level, pairwise constraints for partitional and hierarchical clustering. The book then explores other types of constraints for clustering, including cluster size balancing, minimum cluster size,and cluster-level relational constraints.
It also describes variations of the traditional clustering under constraints problem as well as approximation algorithms with helpful performance guarantees.
The book ends by applying clustering with constraints to relational data, privacy-preserving data publishing, and video surveillance data. It discusses an interactive visual clustering approach, a distance metric learning approach, existential constraints, and automatically generated constraints.
With contributions from industrial researchers and leading academic experts who pioneered the field, this volume delivers thorough coverage of the capabilities and limitations of constrained clustering methods as well as introduces new types of constraints and clustering algorithms.
Since the publication of the bestselling first edition, many advances have been made in exploratory data analysis (EDA). Covering innovative approaches for dimensionality reduction, clustering, and visualization, Exploratory Data Analysis with MATLAB®, Second Edition uses numerous examples and applications to show how the methods are used in practice.
New to the Second Edition
Discussions of nonnegative matrix factorization, linear discriminant analysis, curvilinear component analysis, independent component analysis, and smoothing splines An expanded set of methods for estimating the intrinsic dimensionality of a data set Several clustering methods, including probabilistic latent semantic analysis and spectral-based clustering Additional visualization methods, such as a rangefinder boxplot, scatterplots with marginal histograms, biplots, and a new method called Andrews’ images Instructions on a free MATLAB GUI toolbox for EDA
Like its predecessor, this edition continues to focus on using EDA methods, rather than theoretical aspects. The MATLAB codes for the examples, EDA toolboxes, data sets, and color versions of all figures are available for download at http://pi-sigma.info
Processing multimedia content has emerged as a key area for the application of machine learning techniques, where the objectives are to provide insight into the domain from which the data is drawn, and to organize that data and improve the performance of the processes manipulating it. Applying machine learning techniques to multimedia content involves special considerations – the data is typically of very high dimension, and the normal distinction between supervised and unsupervised techniques does not always apply.
This book provides a comprehensive coverage of the most important machine learning techniques used and their application in this domain. Arising from the EU MUSCLE network, a program that drew together multidisciplinary teams with expertise in machine learning, pattern recognition, artificial intelligence, and image, video, text and crossmedia processing, the book first introduces the machine learning principles and techniques that are applied in multimedia data processing and analysis. The second part focuses on multimedia data processing applications, with chapters examining specific machine learning issues in domains such as image retrieval, biometrics, semantic labelling, mobile devices, and mining in text and music.
This book will be suitable for practitioners, researchers and students engaged with machine learning in multimedia applications.
This book presents important recent developments in mathematical and computational methods used in impedance imaging and the theory of composite materials. By augmenting the theory with interesting practical examples and numerical illustrations, the exposition brings simplicity to the advanced material. An introductory chapter covers the necessary basics. An extensive bibliography and open problems at the end of each chapter enhance the text.