ADVANCES IN MACHINE LEARNING RESEARCH

ADVANCES IN MACHINE LEARNING RESEARCH

Library of Congress Cataloging-in-Publication Data

Contents

Preface

Chapter 1: Enhancing Document Search with a Dynamic Artificial Neural Network

Abstract

1.0. Introduction

2.0. Background

3.0. A Dynamic Architecture

for Artificial Neural Networks

3.1. DAN2 for Text Classification

3.2. Starting Point

3.3. Scalability

3.4. Hierarchical DAN2 for Multiple Class Classification

4.0. Performance Assessment of DAN2

on a Text Classification Benchmark

4.1. The Reuters Benchmark Dataset

4.2. Computational Strategies, Requirements, and Environment

4.3. Performance Measurement Metrics

4.4. DAN2 Results for the Reuters-21578

4.5. DAN2 Computational Times for Reuters-21578

Conclusion

References

Chapter 2: Combination of Depth and Texture Descriptors for Gesture Recognition

Abstract

1. Introduction

2. Method Overview

2.1. Reprojection

2.2. Hand Segmentation and Palm Recognition

2.3. Color Space Transformation

2.4. Geometric Feature Extraction

2.4.1. Distance Features

2.2.2. Curvature Features

2.2.3. Elevation Features

2.2.4. Palm Area Features

2.5. Texture Feature Extraction

2.5.1. Extracting Texture Feature from Curvature

2.5.2. Local Phase Quantization

2.5.3. Histogram of Gradients

2.5.4. Local Ternary Patterns

2.6. Classification

2.6.1. Random Subspace Ensemble of Support Vector Machines

2.6.2. Random Subspace Ensemble of Rotboost with NPE (RSR)

3. Experimental Results

Conclusion

References

Chapter 3: Optimization for Multi Layer Perceptron: Without the Gradient

Abstract

1. Introduction

2. Gradient Learning

3. Bipropagation Algorithm

3.1. Description of the Bipropagation Algorithm

4. Border Pairs Method

4.1. Border Pairs Definition

4.2. The Impact of the Learning Data Feature on Border Pairs

4.2.1. Influence of the Number of Patterns on the Number

of Border Pairs

4.2.2. The Impact of the Ratio of Learning Patterns on the Number of Border Pairs

4.2.3. The Impact of Noise on the Number of Border Pairs

4.2.4. The Impact of Noise on Outliers

5. Noise Reduction with Border Pairs

5.1. The Principle of Noise Reduction with Border Pairs

6. Clustering Data with the Border Pairs Method

6.1. Clustering with Border Pairs

6.2. Complication in Clustering With Border Pairs

6.3. Combining of Border Pairs

7. Classification of Data with the Border

Pairs Method

7.1. Description of the Data Classification with Border

Pairs Method

7.2. Examples of Learning with the Border Pairs Method

7.2.1. XOR

7.2.2. Triangle

7.2.3. Recognition of Irises

7.2.4. Pen-Based Recognition of Handwritten Digits

7.2.5. Ionosphere

7.3. Noisy Data

8. Dynamic Learning with Border Pairs Method

8.1. Dynamic Learning Approaches with Border Pairs Method

8.2. Incremental Learning with the Border Pairs Method

8.3. Online Learning with a Method of Border Pairs

8.3.1. Online Recognition of Digits

Conclusion

References

Chapter 4: Prediction of Cyanotoxin Production along with Cyanobacteria Presence Using Genetic Algorithms and Multivariate Adaptive Regression Splines

Abstract

1. Introduction

2. Experimental Data Set

3. Mathematical Model

3.1. Genetic Algorithms

3.2. Multivariate Adaptive Regression Splines

4. Results

5. Discussion and Conclusion

Acknowledgments

Appendix A

References

Editor Contact Information

Index