Support vector machines : evolution and applications

"Support Vector Machines: Evolution and Applications reviews the basics of Support Vector Machines (SVM), their evolution and applications in diverse fields. SVM is an efficient supervised learning approach popularly used for pattern recognition, medical image classification, face recognition and various other applications. In the last 25 years, a lot of research has been carried out to extend the use of SVM to a variety of domains. This book is an attempt to present the description of a conventional SVM, along with discussion of its different versions and recent application areas. The first chapter of this book introduces SVM and presents the optimization problems for a conventional SVM. Another chapter discusses the journey of SVM over a period of more than two decades. SVM is proposed as a separating hyperplane classifier that partitions the data belonging to two classes. Later on, various versions of SVM are proposed that obtain two hyperplanes instead of one. A few of these variants of SVM are discussed in this book. The major part of this book discusses some interesting applications of SVM in areas like quantitative diagnosis of rotor vibration process faults through power spectrum entropy-based SVM, hardware architectures of SVM applied in pattern recognition systems, speaker recognition using SVM, classification of iron ore in mines and simultaneous prediction of the density and viscosity for the ternary system water- ethanol-ethylene glycol ionic liquids. The latter part of the book is dedicated to various approaches for the extension of SVM and similar classifiers to a multi-category framework, so that they can be used for the classification of data with more than two classes"-- Provided by publisher Contents Preface Acknowledgments List of Acronyms Chapter 1 Introduction to Support Vector Machines Chapter 2 Journey of Support Vector Machines: From Maximum-Margin Hyperplane to a Pair of Non-Parallel Hyperplanes Chapter 3 Power Spectrum Entropy-Based Support Vector Machine for Quantitative Diagnosis of Rotor Vibration Process Faults Abstract Introduction Process Power Spectrum Entropy Method Support Vector Machine Method Fundamental Theory SVM Multiclass Classification Method Rotor Simulation Experiment Selection of Typical Faults Experiment of Simulation Test Rig Experimental Process Original Data Analysis of Fault Data Process diagnosis of Rotor Vibration Faults Establishment of PPSE-SVM Model Extraction of PPSE Feature and Selection of Parameter Diagnosis of Rotor Vibration Fault Category Diagnosis of Fault Degree Diagnosis of Fault Points Verification of PPSE-SVM Robustness Conclusion Acknowledgment References Biographical Sketch Chapter 4 Hardware Architectures of Support Vector Machine Applied in Pattern Recognition Systems Abstract Introduction Related Works Literature Review Low-Power Hardware Implementation on FPGA of SVM for Neural Seizure Detection A Cascade SVM Classifier on FPGA for Early Melanoma Detection Multi-Class SVM Classification for Epilepsy and Epileptic Seizure Detection on VLSI Technology CNN-SVM Hybrid Classifier in Hardware Implementation on FPGA State-of-the-Art Review of SVM Hardware Implementations in FPGA Combinatorial Circuit Application of SVM in FPGA Implementation Blocks of Combinatorial Circuits with an FSM of SVM in FPGA Implementation Synchronous Hardware Architectures of SVM Classifier Applied in an ASR System Speech Signal Processing Training Phase: PSO-SVM Hybrid Algorithm Classification Phase: Synchronous and Asynchronous Hardware Architectures Architecture I: Combinatorial Circuit Architecture II: Pipeline Datapath Controlled by FSM Architecture III: Linear Pipeline with Synchronous Control Architecture IV: Asynchronous Hardware Architecture of SVM Classifier Applied in an ASR System Asynchronous Paradigm: Some Concepts Hardware Description: Linear Pipeline with Asynchronous Control Conclusion and Future Work References Biographical Sketches Chapter 5 Speaker Recognition Using Support Vector Machine Abstract Introduction Speech Corpus Corpus 1 Corpus 2 Feature Extraction and Classification Mel Frequency Cepstral Coefficient Support Vector Machine Model Evaluation Using Confusion Matrix Performance Parameters Proposed Algorithm Experiments and Results Experiment on Corpus 1 Experiment 1: Trained and Tested on Speech Samples Collected from Device 0 Experiment Performed on Corpus 2 Experiment: To Study the Impact of the Number of Utterances on the Accuracy of the System Experiment: Impact of Recording Devices and Recording Languages Conclusion and Future Scope References Biographical Sketches Chapter 6 Application of Support Vector Machine (SVM) in Classification of Iron Ores in Mines Abstract Introduction Methods Sample Collection Image Acquisition Feature Extraction Feature Dimension Reduction Laboratory Analysis Model Development Performance Evaluation Conclusion References Biographical Sketches Chapter 7 Multi-Category Classification Chapter 8 Simultaneous Prediction of the Density and Viscosity for the Ternary System Water-Ethanol–Ethylene Glycol Ionic Liquids Using Support Vector Machine Abstract Introduction Experimental Methodology Validation Results and Discussion Conclusion Acknowledgment References About the Editor Index Blank Page Blank Page