Supervised and Unsupervised Pattern Recognition: Feature Extraction and Computational Intelligence

ISBN-13: 9780849322785 / Angielski / Twarda / 1999 / 388 str.

Evangelia Micheli-Tzanakou; E. M. Tzanakou

Opis

There are many books on neural networks, some of which cover computati onal intelligence, but none that incorporate both feature extraction a nd computational intelligence, as Supervised and Unsupervised Pattern Recognition does. This volume describes the application of a novel, un supervised pattern recognition scheme to the classification of various types of waveforms and images.

Recenzje

"This book is an excellent source of knowledge of state-of-the-art feature extraction...Supervised and unsupervised learning and training schemes are notable finds...Exciting applications of signal and image analysis and recognition...This book provides in-depth guidance and inspiring ideas to new applications of signal and image analysis and recognition."
--Tonglei Li, Ph.D., Purdue University, School of Pharmacy
"...great efforts have been made in a number of communities to explore solutions to pattern recognition problems...this book describes their efforts made over ten researchers in the Neuroelectric and Neurocomputing Laboratories at Rutgers University. Along with concise introductory materials in pattern recognition, this volume presents several applications of supervised and unsupervised schemes to the classification of various types of signals and images...Unlike other books in neural networks, this book gives an emphasis on feature extraction as well, which provides a systematic way to deal with pattern recognition problems in terms of neural networks and computational intelligence...it is worth noting that each chapter contains an extensive bibliography that provides a reliable list of good references. We believe that readers will find this list very useful to understand the materials in the book and cautious beginners in the related fields might benefit from this list as well...helpful to a broad audience of graduate students, researchers, practicing engineers and professionals in computer and information science, electrical engineering, and biomedical informatics...this book reflects the long-term continuous endeavors of a research group for conducting innovatory researches, which could provide some useful hints to those novices in related fields...pioneering volume...welcomed by all interested in the fields of pattern recognition and computational intelligence...the editor's serious attempt to address the aforementioned issue must be welcomed by all interested in the fields of pattern recognition and computational intelligence and, therefore, this book deserves all credit."
--Ke Chen, National Laboratory of Machine Perception and The Center for Information Science, Peking University, Beijing, China
Promo Copy

Spis treści

classifiers-an overview
Criteria for optimal classifier design
Categorizing the Classifiers
Classifiers
Neural Networks
Comparison of Experimental Results
System Performance Assessment
Analysis of Prediction Rates from Bootstrapping Assessment
ARTIFICIAL NEURAL NETWORKS: DEFINITIONS, METHODS, APPLICATIONS
Definitions
Training Algorithm
Some Applications
A SYSTEM FOR HANDWRITTEN DIGIT RECOGNITION
Preprocessing of Handwritten Digit Images
Zernike Moments (ZM) for Characterization of Image Patterns
Dimensionality Reduction
Analysis of Prediction Error Rates from Bootstrapping Assessment
Summary
OTHER TYPES OF FEATURE EXTRACTION METHODS
Introduction
Wavelets
Invariant Moments
Entropy
Cepstrum Analysis
Fractal Dimension
Entropy
SGLD Texture Features
FUZZY NEURAL NETWORKS
Pattern Recognition
Optimization
System Design
Clustering
APPLICATION TO HANDWRITTEN DIGITS
Introduction to Character Recognition
Data Collection
Results
Discussion
Summary
A UNSUPERVISED NEURAL NETWORK SYSTEM FOR VISUAL EVOKED POTENTIALS
Data Collection and Preprocessing
System Design
Results
Discussion
CLASSIFICATION OF MAMMOGRAMS USING A MODULAR NEURAL NETWORK
Methods and System Overview
Modular Neural Networks
Neural Network Training
Classification Results
The Process of Obtaining Results
ALOPEX Parameters
Generalization
Conclusions
"VISUAL OPHTHALMOLOGIST": AN AUTOMATED SYSTEM FOR CLASSIFICATION OF RETINAL DAMAGE
System Overview
Modular Neural Networks
Applications to Ophthalmology
Results
Discussion
A THREE-DIMENSIONAL NEURAL NETWORK ARCHITECTURE
The Neural Network Architecture
Simulations
Discussion
A FEATURE EXTRACTION ALGORITHM USING CONNECTIVITY STRENGTHS AND MOMENT INVARIANTS
ALOPEX Algorithms
Moment Invariants and ALOPEX
Results and Discussion
MULTILAYER PERCEPTRONS WITH ALOPEX: 2D-TEMPLATE MATCHING AND VLSI IMPLEMENTATION
Multilayer Perceptron and Template Matching
VLSI Implementation of ALOPEX
IMPLEMENTING NEURAL NETWORKS IN SILICON
The Living Neuron
Neuromorphic Models
Neurological Process Modeling
SPEAKER IDENTIFICATION THROUGH WAVELET MULTIRESOLUTION DECOMPOSITION AND ALOPEX
Multiresolution Analysis through Wavelet Decomposition
Pattern Recognition with ALOPEX
Methods
Results
Discussion
FACE RECOGNITION IN ALZHEIMER'S DISEASE: A SIMULATION
Methods
Results
Discussion
SELF-LEARNING LAYERED NEURAL NETWORKS
Neocognition and Pattern Classification
Objectives
Methods
Study A
Study B
Summary and Discussion
BIOLOGICAL AND MACHINE VISION
Distributed Representation
The Model
A Modified ALOPEX Algorithm
Application to Template Matching
Brain-to-Computer Link
Discussion
Each section also has an introduction and references

O autorze

Evangelia Miche Tzanakou