Signal Processing Techniques for Knowledge Extraction and Information Fusion

by Mandic, Danilo P.; Golz, Martin; Kuh, Anthony; Obradovic, Dragan; Tanaka, Toshihisa

• ISBN13: 9780387743660
• ISBN10: 0387743669
• Format: Hardcover
• Copyright: 2008-03-01
• Publisher: Springer Verlag

Summary

This state-of-the-art resource brings together the latest findings from the cross-fertilization of signal processing, machine learning and computer science. The emphasis is on demonstrating synergy of different signal processing methods with knowledge extraction and heterogeneous information fusion. Issues related to the processing of signals with low signal-to-noise ratio, solving real-world multi-channel problems, and using adaptive techniques where nonstationarity, uncertainty and complexity play major roles are addressed. Particular methods include Independent Component Analysis, Support Vector Machines, Distributed and Collaborative Adaptive Filtering, Empirical Mode Decomposition, Self Organizing Maps, Fuzzy Logic, Evolutionary Algorithms and several others used frequently in these fields. Also included are both important and novel applications from telecommunications, renewable energy and biomedical engineering.

Table of Contents

Collaborative Signal Processing Algorithms
Collaborative Adaptive Filters for Online Knowledge Extraction and Information Fusion	p. 3
Introduction	p. 3
Previous Online Approaches	p. 5
Collaborative Adaptive Filters	p. 6
Derivation of The Hybrid Filter	p. 7
Detection of the Nature of Signals: Nonlinearity	p. 8
Tracking Changes in Nonlinearity of Signals	p. 10
Detection of the Nature of Signals: Complex Domain	p. 12
Split-Complex vs. Fully-Complex	p. 13
Complex Nature of Wind	p. 17
Conclusions	p. 19
References	p. 20
Wind Modelling and its Possible Application to Control of Wind Farms	p. 23
Formulating Yaw Control for a Wind Turbine	p. 23
Characteristics for Time Series of the Wind	p. 25
Surrogate Data	p. 25
Results	p. 25
Modelling and Predicting the Wind	p. 27
Multivariate Embedding	p. 27
Radial Basis Functions	p. 28
Possible Coordinate Systems	p. 30
Direct vs. Iterative Methods	p. 30
Measurements of the Wind	p. 30
Results	p. 32
Applying the Wind Prediction to the Yaw Control	p. 34
Conclusions	p. 34
References	p. 35
Hierarchical Filters in a Collaborative Filtering Framework for System Identification and Knowledge Retrieval	p. 37
Introduction	p. 37
Hierarchical Structures	p. 39
Generalised Structures	p. 40
Equivalence with FIR	p. 41
Multilayer Adaptive Algorithms	p. 43
The Hierarchical Least Mean Square Algorithm	p. 43
Evaluation of the Performance of HLMS	p. 44
The Hierarchical Gradient Descent Algorithm	p. 45
Applications	p. 46
Standard Filtering Applications	p. 46
Knowledge Extraction	p. 47
Conclusions	p. 49
Mathematical Analysis of the HLMS	p. 50
References	p. 53
Acoustic Parameter Extraction From Occupied Rooms Utilizing Blind Source Separation	p. 55
Introduction	p. 55
Blind Estimation of Room RT in Occupied Rooms	p. 57
MLE-Based RT Estimation Method	p. 57
Proposed Noise Reducing Preprocessing	p. 59
A Demonstrative Study	p. 60
Blind Source Separation	p. 62
Adaptive Noise Cancellation	p. 65
Simulation Results	p. 67
Discussion	p. 72
Conclusion	p. 73
References	p. 74
Signal Processing for Source Localization
Sensor Network Localization Using Least Squares Kernel Regression	p. 77
Introduction	p. 77
Sensor Network Model	p. 80
Localization Using Classification Methods	p. 81
Least Squares Subspace Kernel Regression Algorithm	p. 82
Least Squares Kernel Subspace Algorithm	p. 82
Recursive Kernel Subspace Least Squares Algorithm	p. 84
Localization Using Kernel Regression Algorithms	p. 85
Centralized Kernel Regression	p. 85
Kernel Regression for Mobile Sensors	p. 86
Distributed Kernel Regression	p. 87
Simulations	p. 89
Stationary Motes	p. 90
Mobile Motes	p. 91
Distributed Algorithm	p. 92
Summary and Further Directions	p. 93
References	p. 94
Adaptive Localization in Wireless Networks	p. 97
Introduction	p. 97
RF Propagation Modelling	p. 98
Characteristics of the Indoor Propagation Channel	p. 99
Parametric Channel Models	p. 99
Geo Map-Based Models	p. 100
Non-Parametric Models	p. 102
Localization Solution	p. 103
Simultaneous Localization and Learning	p. 104
Kohonen SOM	p. 105
Main Algorithm	p. 106
Comparison Between SOM and SLL	p. 107
Convergence Properties of SLL	p. 107
Statistical Conditions for SLL	p. 113
Results on 2D Real-World Scenarios	p. 116
Conclusions	p. 118
References	p. 119
Signal Processing Methods for Doppler Radar Heart Rate Monitoring	p. 121
Introduction	p. 121
Signal Model	p. 123
Physiological Signal Model	p. 125
Single Person Signal Processing	p. 126
Demodulation	p. 126
Detection of Heartbeat and Estimation of Heart Rate	p. 127
Multiple People Signal Processing	p. 132
Heartbeat Signal	p. 133
Algorithm	p. 133
Results	p. 134
Conclusion	p. 138
References	p. 139
Multimodal Fusion for Car Navigation Systems	p. 141
Introduction	p. 141
Kalman Filter-Based Sensor Fusion for Dead Reckoning Improvement	p. 143
Map Matching Improvement by Pattern Recognition	p. 146
Generation of Feature Vectors by State Machines	p. 147
Evaluation of Certainties of Road Alternatives Based on Feature Vector Comparison	p. 150
Fuzzy Guidance	p. 154
Conclusions	p. 157
References	p. 157
Information Fusion in Imaging
Cue and Sensor Fusion for Independent Moving Objects Detection and Description in Driving Scenes	p. 161
Introduction	p. 161
Vision Sensor Data Processing	p. 164
Vision Sensor Setup	p. 164
Independent Motion Stream	p. 165
Recognition Stream	p. 167
Training	p. 168
Visual Streams Fusion	p. 170
IMO Detection and Tracking	p. 171
Classification and Description of the IMOs	p. 171
LIDAR Sensor Data Processing	p. 172
LIDAR Sensor Setup	p. 172
Ground Plane Estimation	p. 173
LIDAR Obstacles Projection	p. 175
Vision and LIDAR Fusion	p. 175
Results	p. 176
Conclusions and Future Steps	p. 177
References	p. 178
Distributed Vision Networks for Human Pose Analysis	p. 181
Introduction	p. 181
A Unifying Framework	p. 183
Smart Camera Networks	p. 184
Opportunistic Fusion Mechanisms	p. 185
Human Posture Estimation	p. 187
The 3D Human Body Model	p. 189
In-Node Feature Extraction	p. 190
Collaborative Posture Estimation	p. 192
Towards Behavior Interpretation	p. 195
Conclusions	p. 198
References	p. 199
Skin Color Separation and Synthesis for E-Cosmetics	p. 201
Introduction	p. 201
Image-Based Skin Color Analysis and Synthesis	p. 203
Shading Removal by Color Vector Space Analysis: Simple Inverse Lighting Technique	p. 205
Imaging Model	p. 205
Finding the Skin Color Plane in the Face and Projection Technique for Shading Removal	p. 208
Validation of the Analysis	p. 210
Image-Based Skin Color and Texture Analysis/Synthesis	p. 211
Data-Driven Physiologically Based Skin Texture Control	p. 212
Conclusion and Discussion	p. 218
References	p. 219
ICA for Fusion of Brain Imaging Data	p. 221
Introduction	p. 221
An Overview of Different Approaches for Fusion	p. 223
A Brief Description of Imaging Modalities and Feature Generation	p. 224
Functional Magnetic Resonance Imaging	p. 224
Structural Magnetic Resonance Imaging	p. 226
Diffusion Tensor Imaging	p. 226
Electroencephalogram	p. 227
Brain Imaging Feature Generation	p. 228
Feature-Based Fusion Framework Using ICA	p. 228
Application of the Fusion Framework	p. 230
Multitask fMRI	p. 231
Functional Magnetic Resonance Imaging-Structural Functional Magnetic Resonance Imaging	p. 231
Functional Magnetic Resonance Imaging-Event-Related Potential	p. 233
Structural Magnetic Resonance Imaging-Diffusion Tensor Imaging	p. 233
Parallel Independent Component Analysis	p. 235
Selection of Joint Components	p. 235
Conclusion	p. 237
References	p. 237
Knowledge Extraction in Brain Science
Complex Empirical Mode Decomposition for Multichannel Information Fusion	p. 243
Introduction	p. 243
Data Fusion Principles	p. 244
Empirical Mode Decomposition	p. 244
Ensemble Empirical Mode Decomposition	p. 247
Extending EMD to the Complex Domain	p. 249
Complex Empirical Mode Decomposition	p. 251
Rotation Invariant Empirical Mode Decomposition	p. 254
Complex EMD as Knowledge Extraction Tool for Brain Prosthetics	p. 254
Empirical Mode Decomposition as a Fixed Point Iteration	p. 257
Discussion and Conclusions	p. 258
References	p. 259
Information Fusion for Perceptual Feedback: A Brain Activity Sonification Approach	p. 261
Introduction	p. 261
Principles of Brain Sonification	p. 263
Empirical Mode Decomposition	p. 264
EEG and EMD: A Match Made in Heaven?	p. 265
Time-Frequency Analysis of EEG and MIDI Representation	p. 269
Experiments	p. 271
Conclusions	p. 272
References	p. 273
Advanced EEG Signal Processing in Brain Death Diagnosis	p. 275
Introduction	p. 275
Background and EEG Recordings	p. 276
Diagnosis of Brain Death	p. 276
EEG Preliminary Examination and Diagnosis System	p. 276
EEG Recordings	p. 278
EEG Signal Processing	p. 279
A Model of EEG Signal Analysis	p. 280
A Robust Prewhitening Method for Noise Reduction	p. 280
Independent Component Analysis	p. 283
Fourier Analysis and Time-Frequency Analysis	p. 285
EEG Preliminary Examination with ICA	p. 285
Extracted EEG Brain Activity from Comatose Patients	p. 286
The Patients Without EEG Brain Activities	p. 287
Quantitative EEG Analysis with Complexity Measures	p. 288
The Approximate Entropy	p. 289
The Normalized Singular Spectrum Entropy	p. 290
The C[subscript 0] Complexity	p. 291
Detrended Fluctuation Analysis	p. 292
Quantitative Comparison Results	p. 292
Classification	p. 295
Conclusion and Future Study	p. 296
References	p. 297
Automatic Knowledge Extraction: Fusion of Human Expert Ratings and Biosignal Features for Fatigue Monitoring Applications	p. 299
Introduction	p. 299
Fatigue Monitoring	p. 301
Problem	p. 301
Human Expert Ratings	p. 302
Experiments	p. 303
Feature Extraction	p. 305
Feature Fusion and Classification	p. 306
Learning Vector Quantization	p. 307
Automatic Relevance Determination	p. 308
Support Vector Machines	p. 309
Results	p. 310
Feature Fusion	p. 310
Feature Relevance	p. 312
Intra-Subject and Inter-Subject Variability	p. 313
Conclusions and Future Work	p. 314
References	p. 315
Index	p. 317
Table of Contents provided by Ingram. All Rights Reserved.