Advances in Machine Learning and Data Analysis (eBook)

Preface 
6 
Contents 7
1 2D/3D Image Data Analysis for Object Tracking and Classification 9
1.1 Introduction 9
1.2 2D/3D Vision System 10
1.2.1 3D-Time of Flight Camera 11
1.2.2 2D/3D Image Registration and Synchronization 12
1.2.2.1 Temporal Synchronization 12
1.2.2.2 Spatial Registration 13
1.3 Multimodal Data Fusion and Segmentation 15
1.4 Object Tracking and Classification 15
1.5 Real Time Hand Based Robot Control Using 2D/3D Images 18
1.5.1 Set-Up 18
1.5.2 Control Application 19
1.5.3 Experimental Results 20
1.6 Conclusion 20
References 20
2 Robot Competence Development by Constructive Learning 22
2.1 Introduction 22
2.2 Sensory-Motor Mapping Development Via Constructive Learning 24
2.2.1 Why Constructive Learning? 24
2.2.2 Topological Development of the Sensory-Motor Mapping Network 25
2.2.3 Parameter Adjustments of the Sensory-Motor Mapping Network 25
2.3 Adaptation of Sensory-Motor Mapping 26
2.4 Experimental Studies 28
2.4.1 Experimental System 28
2.4.2 Constructive Learning and Adaptation in Tool-Use 29
2.5 Conclusions 31
References 31
3 Using Digital Watermarking for Securing Next Generation Media Broadcasts 34
3.1 Introduction 34
3.2 Framework Overview 35
3.3 PKI Framework 37
3.3.1 Overview 37
3.4 Watermarking Framework 38
3.4.1 Overview 38
3.4.2 Signature Schemes 40
3.4.2.1 The Nyberg-Rueppel Signature Scheme 40
3.4.2.2 Short Hash Methods 40
3.4.2.3 Hash Table for Verification Purposes 41
3.4.2.4 Case Study: Video Broadcaster 42
3.4.3 Suitable Watermarking Algorithms 43
3.4.3.1 Proposed Watermarking Algorithm 44
3.4.4 Embedding the Watermark 44
3.4.5 Retrieval of the Watermark 45
3.4.6 Verifying the Signature 46
3.5 Conclusions and Future Work 47
References 47
4 A Reduced-Dimension Processor Model 49
4.1 Introduction 49
4.1.1 Paper Outline 50
4.1.2 Related Research 50
4.1.3 A Brief Introduction to Artificial Neural Networks 51
4.1.4 An Overview of Principal Component Analysis 52
4.2 Experimental Setup and Data Processing 52
4.2.1 Data Acquisition 52
4.2.2 Dimension Reduction 54
4.3 Neural Network Structure and Training 58
4.4 Processor Performance Prediction with the Neural Network Model 60
4.5 Conclusions 61
References 61
5 Hybrid Machine Learning Model for Continuous Microarray Time Series 63
5.1 Introduction 63
5.1.1 Computational Methods for Microarray Time Series Analysis 64
5.2 Machine Learning Methods 65
5.2.1 Neural Network Models 66
5.2.2 Neural Network Models for Microarray Analysis 66
5.2.3 Dimension Reduction and Transformation 68
5.2.4 Principal Component Analysis 68
5.2.5 Independent Component Analysis 69
5.3 The Proposed Hybrid PCA-NN Machine Learning Model 70
5.4 The Microarray Time Series Datasets 73
5.5 Experimental Results 73
5.5.1 Models with Stand-Alone Neural Network 73
5.5.2 Hybrid Algorithms of Principal Component and Neural Network 74
5.5.3 Results Comparison: Hybrid PCA-NN Models' Performance and Other Existing Algorithms 76
5.5.4 Analysis on the Network Structure and the Out-of-Sample Validations 77
5.6 Result Discussions and Conclusion 80
References 80
6 An Asymptotic Method to a Financial Optimization Problem 84
6.1 Introduction 84
6.2 Integral Representation of the Solution 86
6.3 Properties of the Free Boundary 87
6.4 Numerical Solution of the Free Boundary 92
6.5 Asymptotic Analysis of the Free Boundary 93
6.6 Global Approximation Formulas 97
6.7 Conclusion 98
References 98
7 Analytical Design of Robust Multi-loop PI Controller for Multi-time Delay Processes 100
7.1 Introduction 100
7.2 The Multi-loop Feedback Controller Design for Desired Set-Point Responses 102
7.3 Reduction to the Multi-loop PI Controller 104
7.4 Example of Two-Input, Two-Output (TITO) Case 105
7.5 Robust Stability Analysis 106
7.6 Simulation Study 107
7.7 Conclusions 112
References 112
8 Automatic and Semi-automatic Methods for the Detection of Quasars in Sky Surveys 114
8.1 Introduction 114
8.1.1 Variability Properties of the Quasar Light Curves 115
8.2 Automatic and Semi-automatic Methods for Quasar Detection 117
8.2.1 Variability Selection Method for Quasar Candidate in MACHO 117
Data: Deviation from a constant brightness light curve 118
Step 1: Magnitude and color tests 118
Step 2: Statistics tests of light curve variability 118
Step 3: Removal of periodic variable stars 119
Step 4: Examination by eye 119
8.2.2 Variability Selection Methods for Quasar Candidate in OGLE-II 119
First criterion: selection on photometry: magnitude and colour 120
Second criterion: the slope of variograms 120
Third criterion: QSO and Be star colors 120
Forth criterion: Manual selection 120
Follow-up spectroscopy experiments and works 121
8.2.3 Automatic Photometric Selection of Quasars from the Sloan Digital Sky Survey 122
Follow-up Works to This Automatic Photometric Selection of Quasars 122
Nonparametric Bayesian Classification (NBC) 123
Fast Algorithms for Computing the Kernel Density Estimate 124
8.3 Methodology for the Exploration of the Quasar Autocorrelation 125
8.3.1 Experimental Data and Data Processing 125
8.3.2 Autocorrelation of the Time Series 126
8.3.3 Dynamic Time Warping 126
8.3.4 Hierarchical Clustering of the Autocorrelation Sequences 127
8.4 Exploratory Experimental Results 129
8.4.1 Experimental Results with the Quasar Dataset 129
8.4.2 Experimental Results with the Be Star Dataset 133
8.4.3 Experimental Results with the Combined Quasar and Be Star Dataset 133
References 142
9 Improving Low-Cost Sail Simulator Results by Artificial Neural Networks Models 144
9.1 Introduction 144
9.2 Fluid Dynamics Simulation 146
9.3 Sail Shape Definition 147
9.4 Experimental Validation 149
9.4.1 Experimental Apparatus 149
9.4.2 Sail Shape Analysis 149
9.4.3 Forces Analysis 150
9.5 Neural Network Correction 151
9.5.1 Integrating Computational and Experimental Analyses 151
9.6 Conclusion 153
References 153
10 Rough Set Approaches to Unsupervised Neural Network Based Pattern Classifier 155
10.1 Introduction 155
10.1.1 Basics of Rough Set Theory 156
10.1.1.1 Information System (IS) 156
10.1.1.2 Reducts 157
10.1.1.3 Core 157
10.1.1.4 Discernibility Matrix 157
10.2 Image Processing and Feature Extraction for the First Case Study 157
10.3 Discretization 159
10.3.1 Algorithm 159
10.3.2 Steps to Obtain Cuts 160
10.3.3 Steps for Algorithm 160
10.4 Reduction of Attributes 160
10.4.1 Steps for Finding Reduced Set of Attributes 161
10.4.2 Algorithm for Finding Reducts 161
10.5 Rough Neuro Based Hybrid Approach and the Experimentation Done 161
10.6 Results 163
10.6.1 Results for Case Study-1 164
10.6.2 Results for Case Study-2 165
10.7 Conclusion 167
References 167
11 A New Robust Combined Method for Auto Exposure and Auto White-Balance 168
11.1 Introduction 168
11.2 Auto Exposure Algorithm for Lighting Condition Detection 169
11.2.1 Lighting Condition Detection 169
11.2.2 Auto Exposure 171
11.3 Multiple Exposure Mechanism and Auto White-Balance 173
11.3.1 Multiple Exposure Mechanism 173
11.3.2 Detecting One-Primary Color Images 174
11.3.3 Auto White-Balance in Conjunction with Multiple Exposure 175
11.4 Simulations and Comparisons 177
11.5 Conclusions 179
References 179
12 A Mathematical Analysis Around Capacitive Characteristics of the Current of CSCT: Optimum Utilization of Capacitors of Harmonic Filters 182
12.1 Introduction 182
12.2 Mathematical Analysis 184
12.3 Conclusion 193
References 193
13 Harmonic Analysis and Optimum Allocation of Filters in CSCT 194
13.1 Introduction 194
13.2 Mathematical Analysis to Calculate Harmonic Components of the Current 195
13.3 Conclusion 205
References 205
14 Digital Pen and Paper Technology as a Means of Classroom Administration Relief 206
14.1 Introduction 206
14.1.1 Attendance Tracking and Grading in German Schools 208
14.1.2 Special Case: The International School of Bremen 209
14.1.3 Digital Pen and Paper Technology 209
14.1.4 The MTT System 210
14.1.5 Previous Studies on the Digital Pen and Paper Technology 210
14.2 Methodology 211
14.2.1 Practical Test and Observation 211
14.2.2 Interviews 213
14.2.3 Form Analysis 213
14.3 Results and Discussion 213
14.3.1 Viability of the Digital Pen in a Classroom Environment 213
14.3.2 Viability of Static Forms in a Classroom Environment 214
14.3.3 Viability of Text Recognition 215
14.3.4 Benefits of the MTT System in a Classroom Environment 215
14.4 Conclusion 217
References 217
15 A Conceptual Model for a Network-Based Assessment Security System 219
15.1 Introduction 219
15.2 Securing Technology-Enhanced Assessment Environments 221
15.2.1 Secured Testing Environment Solutions 222
15.2.2 Video Monitoring Solutions 223
15.3 The Virtual Invigilator System 223
15.3.1 Architecture 225
15.3.2 Security Features 226
15.3.3 System Design 227
15.3.4 Technical Design – Prototype 229
15.3.5 Exam Rules 230
15.4 Future Enhancements 231
References 231
16 Incorrect Weighting of Absolute Performance in Self-Assessment 233
16.1 Introduction 233
16.2 The Psychology of Self-Evaluation 234
16.3 Experiment 236
16.3.1 Method 236
16.3.2 Results 237
16.3.3 Discussion and Conclusions 239
16.4 Conclusion 241
References 241