Engineering Applications of Neural Networks (eBook)

11th International Conference, EANN 2009, London, UK, August 27-29, 2009, Proceedings

Dominic Palmer-Brown, Chrisina Draganova, Elias Pimenidis, Haris Mouratidis (Autoren)

eBook Download: PDF

2009 | 1. Auflage
XIII, 508 Seiten
Springer-Verlag
978-3-642-03969-0 (ISBN)

Infos

This book constitutes the refereed proceedings of the 11th International Conference on Engineering Applications of Neural Networks, EANN 2009, held in London, GB, on August 27-29, 2009. The 47 revised full papers were carefully reviewed and selected from many submissions. The following diverse topics were discussed: reducing urban concentration, entropy topography in epileptic electroencephalography, phytoplanktonic species recognition, revealing the structure of childhood abdominal pain data, robot control, discriminating angry and happy facial expressions, flood forecasting and assessing credit worthiness.

Inhalt

Preface 5
Organization 6
Table of Contents 8
Intelligent Agents Networks Employing Hybrid Reasoning: Application in Air Quality Monitoring and Improvement 13
Introduction 13
Aim of This Project 13
Air Pollution 14
Theoretical Background 14
Risk Evaluation Using Fuzzy Logic 15
The Agent-Based System 16
System’s Architecture 16
Sensor Agents Architecture 17
Evaluation Agents 19
System’s Agents 21
Decision Agents 23
Actuators 24
Graphical User Interface 24
Pilot Application of the System 25
Running the Agent Network for Heat Index 25
Running the Agent Network for Air Pollutants 26
Conclusions 26
References 27
Neural Network Based Damage Detection of Dynamically Loaded Structures 29
Introduction 29
Methodology of Damage Detection 30
Artificial Neural Network 32
Stochastic Analysis 32
Software Tools 33
Application – Cantilever Beam 34
Conclusions 37
References 37
Reconstruction of Cross-Sectional Missing Data Using Neural Networks 40
Introduction 40
Developing a New Algorithm 41
The Modified GRNN (Generalized Regression Neural Networks) Algorithm 42
The GMI Algorithm 42
Assessing the New Technique 44
Results and Discussion 45
Summary and Conclusion 45
References 46
Municipal Creditworthiness Modelling by Kernel-Based Approaches with Supervised and Semi-supervised Learning 47
Introduction 47
Municipal Creditworthiness Problem Description 48
Basic Notions of Support Vector Machines and Learning 49
Modelling and Analysis of the Results 52
Modelling by SVMs with Supervised Learning 52
Modelling by Kernel-Based Approaches with Semi-supervised Learning 53
Conclusion 55
References 56
Clustering of Pressure Fluctuation Data Using Self-Organizing Map 57
Introduction 57
Acquisition of Pressure Fluctuation Data 58
Methodology for Clustering by the SOM 59
Batch Self-Organizing Map 59
Procedures of Clustering for Classification of Pressure Fluctuation Data and Operational Conditions 60
Results and Discussions 61
Simulations of Clustering of Operational Conditions 61
Prediction of Dynamic Behavior of Interface Based on Clustering Map 64
Conclusions 65
References 66
Intelligent Fuzzy Reasoning for Flood Risk Estimation in River Evros 67
Introduction 67
Necessity for a New Approach 68
Necessity for Applying Flexible Models 69
The Fuzzy Algebra Model 70
Implementation of the IS 73
Application in the Case of the Flood Risk Problem 73
Discussion – Comparison to Existing Approaches 75
References 77
Fuzzy Logic and Artificial Neural Networks for Advanced Authentication Using Soft Biometric Data 79
Introduction 79
System Architecture 80
Audio Hard Feature Extraction and Authentication 81
Fingerprint Hard Feature Extraction and Authentication 83
Soft-Biometric Feature Extraction 84
Artificial Neural Network-Based Soft-Biometric Feature Scoring 85
Fuzzy Logic-Based Fusion and Authentication 85
Performance Evaluation 88
Embedded Implementation 88
Conclusions 89
References 89
Study of Alpha Peak Fitting by Techniques Based on Neural Networks 91
Introduction 91
Existing Solutions 92
Proposed Solution 93
Method 93
Training Data 93
Network Design 94
Inputs and Output 95
Results 96
Conclusions 97
References 97
Information Enhancement Learning: Local Enhanced Information to Detect the Importance of Input Variables in Competitive Learning 98
Introduction 98
Theory and Computational Methods 99
Enhancement and Relaxation 99
Self-enhancing 100
Collective Enhancement 102
Local Enhancement 103
Results and Discussion 104
Artificial Data 104
Conclusion 107
References 108
Flash Flood Forecasting by Statistical Learning in the Absence of Rainfall Forecast: A Case Study 110
Introduction 110
Problem Statement 111
Flash Flood Forecasting 111
$Gardon d’Anduze$ Flash Floods 112
Noise and Accuracy 113
Model Design 113
Definition of the Model 113
Model Selection 114
Training 115
Regularization 115
Weight Decay 115
Early Stopping 116
Results and Discussion 116
Conclusion 118
References 119
An Improved Algorithm for SVMs Classification of Imbalanced Data Sets 120
Introduction 120
Background 121
Support Vector Machines 121
Related Works 123
Boundary Elimination and Domination Algorithm 124
Experiments and Results 126
Experiment Methodology 126
Results 127
Conclusions 129
References 129
Visualization of MIMO Process Dynamics Using Local Dynamic Modelling with Self Organizing Maps 131
Introduction 131
Local Linear Modelling of Dynamics 132
Clustering Dynamics 132
Local Model Estimation 133
Retrieval 133
Visualization of Dynamics 134
Experimental Results 134
Industrial-Scale 4-Tank Model 134
Experiment Description 136
Model Training and Validation 136
Visualization of MIMO Dynamic Features 138
Conclusion 141
References 141
Data Visualisation and Exploration with Prior Knowledge 143
Introduction 143
Data Exploration 144
Standard GTM 144
Extension to Block GTM 146
Extension of GTM for Missing Data Using EM 147
Stabilising the EM Algorithm 147
Assessing Unsupervised Learning 147
Experiments on Artificial Data 148
Experiments on Geochemical Data 151
Conclusions 152
Future Work 153
References 153
Reducing Urban Concentration Using a Neural Network Model 155
Introduction 155
The Neural Network Model 156
Some Highlights of the Model 158
A Real Example 159
Conclusion 163
References 163
Dissimilarity-Based Classification of Multidimensional Signals by Conjoint Elastic Matching: Application to Phytoplanktonic Species Recognition 165
Introduction 165
Dissimilarity Measure for Multidimensional Signals by Conjoint Elastic Matching 166
Comparison of Two 1D Signals by the Classical Method $Dynamic Time Warping$ 166
Neighborhood Restrictions of DTW Algorithm 168
Dissimilarity Measure of Positive Signals 169
Conjoint Elastic Matching of nD Signals 169
Matching Visualizations 169
Application to the Phytoplanktonic Species Identification 171
Data Presentation 171
Applied Classification Methods 173
Classification Results 174
Conclusion 175
References 176
Revealing the Structure of Childhood Abdominal Pain Data and Supporting Diagnostic Decision Making 177
Introduction 178
Random Forest Implementations 179
Experiment Setup 179
Experimental Results and Discussion 182
Genetic Algorithm Clustering and Genetic Feature Selection 184
Experiment Setup 185
Experimental Results and Discussion 186
Conclusions 187
References 187
Relating Halftone Dot Quality to Paper Surface Topography 190
Introduction 190
Data Acquisition 191
Analysis Methods 192
Self Organizing Maps 192
Clustering 193
Support Vector Machine Classification 193
Analysis of Print Quality 194
Analysis of Topography and Print 195
SVM Classification 197
C-SVC Classification 198
$.$-SVM Classification 198
Class Probabilities 199
Conclusion 199
References 200
Combining GRN Modeling and Demonstration-Based Programming for Robot Control 202
Introduction 202
Developing GRN Controllers 203
RNN-Based Regulatory Model 203
Learning Algorithm for Constructing GRN Controllers 205
Demonstration-Based Programming 206
Experiments and Results 206
Modeling GRNs from Expression Data 207
Learning GRNs for Robot Control 208
Conclusions and Future Work 210
References 211
Discriminating Angry, Happy and Neutral Facial Expression: A Comparison of Computational Models 212
Introduction 212
Background 213
Gabor Filters 213
Curvilinear Component Analysis 215
Intrinsic Dimension 216
Classification Using Support Vector Machines 216
Experiments and Results 217
Conclusions 220
References 220
Modeling and Forecasting CAT and HDD Indices for Weather Derivative Pricing 222
Introduction 222
Modeling Temperature Process 224
Wavelet Neural Networks for Multivariate Process Modeling 225
Modeling and Forecasting CAT and HDD Indices 227
Temperature Derivative Pricing 230
Conclusions 232
References 233
Using the Support Vector Machine as a Classification Method for Software Defect Prediction with Static Code Metrics 235
Introduction 235
Background 236
Static Code Metrics 236
The Support Vector Machine 237
Data 237
Method 239
Data Pre-processing 239
Experimental Design 241
Assessing Performance 242
Results 242
Analysis 243
Conclusion 244
References 244
Appendix 246
Adaptive Electrical Signal Post-processing with Varying Representations in Optical Communication Systems 247
Introduction 247
Background to the Problem 248
Description of the Data 249
Representation of the Data 250
The Discrete Wavelet Transform 251
Independent Component Analysis 252
Method 252
Easy and Hard Cases 252
Visualisation Using PCA 253
Single Layer Neural Network 254
Performance Measures 254
Experiments 255
The First Experiment 255
The Second Experiment 255
The Third Experiment 256
Discussion 256
References 257
Using of Artificial Neural Networks (ANN) for Aircraft Motion Parameters Identification 258
Introduction 258
Longitudinal Forces during Takeoff Run 259
Problem Definition 260
Identification Procedure 261
Identification Results 261
Check Modeling 263
Estimation of Actual Aircraft Braking Characteristics under Different Runway Conditions 265
Further Solutions for ANN-Based Identification Tasks 267
References 268
Ellipse Support Vector Data Description 269
Introduction 269
Support Vector Data Description 271
The Proposed Method 272
$/Phi$ Functions Characteristics 276
Experiments 277
Conclusion 279
References 279
Enhanced Radial Basis Function Neural Network Design Using Parallel Evolutionary Algorithms 281
Introduction 281
State of the Art 282
Method Overview 283
Description of EvRBF 283
Description of Symbiotic_CHC_RBF 284
Description of SymbPar 284
Experiments and Results 286
Conclusions and Future Research 290
References 291
New Aspects of the Elastic Net Algorithm for Cluster Analysis 293
Introduction 293
Some New Aspects on the Elastic Net Algorithm 294
Application to Artificial Created Two-Dimensional Clusters 296
Conclusions 301
References 302
Neural Networks for Forecasting in a Multi-skill Call Centre 303
Introduction 303
Forecasting in Call Centres 304
Forecasting in CCs Using NNs 305
Background 305
Data Set 306
Variables 307
Metrics 307
Adaptations 308
Adaptive Learning Rate 308
Results 309
Analysis 309
Comparative 310
Conclusions 311
References 312
Relational Reinforcement Learning Applied to Appearance-Based Object Recognition 313
Introduction 313
Reinforcement Learning 314
Relational Reinforcement Learning 314
Appearance-Based Modeling 315
Application 318
Conclusion 323
Future Research 323
References 324
Sensitivity Analysis of Forest Fire Risk Factors and Development of a Corresponding Fuzzy Inference System: The Case of Greece 325
Introduction 325
Theoretical Framework 326
Pearson Correlation Coefficients 329
Sensitivity Analysis 330
Analysis Related to the Case of Forest Fire Incidents 330
Analysis Related to the Case of the Burned Area 332
Compatibility of the System’s Output to the Actual Case 334
Conclusions and Discussion 335
References 335
Nonmonotone Learning of Recurrent Neural Networks in Symbolic Sequence Processing Applications 337
Introduction 337
Nonmonotone Training Algorithms 339
Experiments and Results 341
Conclusions 346
References 346
Indirect Adaptive Control Using Hopfield-Based Dynamic Neural Network for SISO Nonlinear Systems 348
Introduction 348
Hopfield-Based Dynamic Neural Model 349
Descriptions of the DNN Model 349
Hopfied-Based DNN Approximator 350
Problem Formulation 351
Design of IACHDNN 352
Simulation Results 359
Conclusions 360
References 361
A Neural Network Computational Model of Visual Selective Attention 362
Introduction 362
Proposed Computational Model of Visual Selective Attention 364
Coincidence Detector Neurons and the Correlation Control Module 365
Simulations and Evaluation of the Model 367
Attentional Blink Explanation – Theory 367
Discussion 369
References 369
Simulation of Large Spiking Neural Networks on Distributed Architectures, The “DAMNED” Simulator 371
MIMD-DM Architectures 372
Architecture of DAMNED 372
Delayed Queues of Events 374
Conservative and Distributed Virtual Clock Handling 375
Configuration of DAMNED and Definition of a SNN 378
Results 378
Conclusion and Future Work 380
References 381
A Neural Network Model for the Critical Frequency of the F2 Ionospheric Layer over Cyprus 383
Introduction 383
Characteristics of the F2 Layer Critical Frequency 384
Model Parameters 386
Experiments and Results 387
Conclusions and Future Work 389
References 389
Dictionary-Based ClassificationModels. Applications for Multichannel Neural Activity Analysis 390
Introduction 390
Material and Methods 392
Animal Training and Behavioral Tasks 392
Chronic Animal Preparation and Neural Ensemble Recording 392
Data Analysis 392
Preprocessing 392
Manual Scatterplot Classification 393
Quantification and Classification of Spike Waveforms 393
Definition of SDC 397
Results 398
Conclusion 399
References 399
Pareto-Based Multi-output Metamodeling with Active Learning 401
Introduction 401
Global Surrogate Modeling 402
Multi-objective Modeling 402
Related Work 403
Problems 404
Analytic Function 404
Low Noise Amplifier (LNA) 405
Experimental Setup 405
SUMO-Toolbox 405
Analytic Function (AF) 406
LNA 406
Results 407
Analytic Function: Use Case 1 407
Analytic Function: Use Case 2 408
LNA: Use Case 1 409
LNA: Use Case 2 410
Conclusion and Future Work 411
References 411
Isolating Stock Prices Variation with Neural Networks 413
Introduction 413
Literature Review 414
Experiments and Results 416
Experimental Methodology 416
Results 417
Conclusions 419
References 420
Evolutionary Ranking on Multiple Word Correction Algorithms Using Neural Network Approach 421
Introduction 421
Typing Correction Functions 422
Word List Neural Network Ranking and Definitions 423
Word List Neural Network Ranking Modelling 425
Conclusion 430
References 430
Application of Neural Fuzzy Controller for Streaming Video over IEEE 802.15.1 431
Introduction 431
Methodology 432
Computer Simulation Results 435
Conclusion 440
References 440
Tracking of the Plasma States in a Nuclear Fusion Device Using SOMs 442
Introduction 442
Data Visualization with SOM 443
Quality Indexes 444
The Data Base Composition 445
Results 446
Conclusions 449
References 449
An Application of the Occam Factor to Model Order Determination 450
Introduction 450
The Bayesian Approach 450
Formulating the Error Term E$_{D}$(w) for the Simple Polynomial Model 451
Deriving a Second Order Approximation for the Log Posterior 452
Calculation of the Evidence P (D|HK) 452
Conclusions and Future Work 455
References 455
Use of Data Mining Techniques for Improved Detection of Breast Cancer with Biofield Diagnostic System 456
Introduction 456
Description of the Datasets 458
Dataset 1 458
Dataset 2 458
Proposed Data Mining Framework 459
Results and Discussion 461
TTSH Dataset – Classification Results 461
US Dataset – Classification Results 462
Comparison of Results and Discussion 462
Conclusion 463
References 464
Clustering of Entropy Topography in Epileptic Electroencephalography 465
Introduction 465
The Electroencephalography and the Neuronal Sources 467
Methodology 468
Entropy to Measure “Order” in the Brain 468
Entropy Topography Clustering 469
Results 470
Data Description 470
Movie Visual Review 470
Electrodes Clustering 470
Conclusions 472
References 473
Riverflow Prediction with Artificial Neural Networks 475
Introduction 475
Daily Riverflow Prediction 476
Mekong River at Pakse Gauging Station, Lao 476
Principal Component Analysis 477
Chao Phraya River at Nakhon Sawan Gauging Station, Thailand 478
Daily Stage Prediction 480
Surma River at Sylhet Gauging Station, Bangladesh 480
Performance Criteria 481
Concluding Remarks 482
References 482
Applying Snap-Drift Neural Network to Trajectory Data to Identify Road Types: Assessing the Effect of Trajectory Variability 484
Introduction 484
Past Work on Trajectory Analysis 485
Snap-Drift Neural Networks 486
The Snap-Drift Algorithm 488
Data Collection 489
Road Design Parameter Derivation from GPS Trajectory Data 490
Trajectory Data Variability Analysis 490
Trajectory Variability Reduction 492
Data Types 492
Results 492
Conclusion 494
References 495
Reputation Prediction in Mobile Ad Hoc Networks Using RBF Neural Networks 497
Introduction 497
Related Work 499
Modeling the Network 500
Simulation Details 501
Simulation Results 503
Practical Considerations 505
Conclusions and Future Work 505
References 506
Author Index 507

Produkt-Details

• Erscheint lt. Verlag: 1.1.2009
• Sprache: englisch
• Themenwelt: Mathematik / Informatik ► Informatik ► Netzwerke
• Themenwelt: Informatik ► Theorie / Studium ► Künstliche Intelligenz / Robotik
• Themenwelt: Technik
• Schlagworte: Artificial Intelligence • biometrics • cluster analysis • Data Mining • data reconstruction • Development • Dynamic Loads • face recognition • fuzzy • Fuzzy Logic • Intelligent Agents • learning • Modeling • programming • quality • Quality assurance • Self-Organizing Maps • Software • Visualization
• ISBN-10: 3-642-03969-3 / 3642039693
• ISBN-13: 978-3-642-03969-0 / 9783642039690