Preface	7
	Contents	10
	List of Contributors	13
	Part I Data Mining	19
	1 Optimization in Reproducing Kernel Hilbert Spacesof Spike Trains	20
	1.1 Introduction	21
	1.2 Some Background on RKHS Theory	22
	1.3 Inner Product for Spike Times	24
	1.4 Inner Product for Spike Trains	25
	1.5 Properties and Estimation of the Memoryless Cross-Intensity Kernel	27
	1.5.1 Properties	27
	1.5.2 Estimation	29
	1.6 Induced RKHS and Congruent Spaces	30
	1.6.1 Space Spanned by Intensity Functions	31
	1.6.2 Induced RKHS	31
	1.6.3 mCI Kernel and the RKHS Induced by	32
	1.6.4 mCI Kernel as a Covariance Kernel	33
	1.7 Principal Component Analysis	34
	1.7.1 Optimization in the RKHS	34
	1.7.2 Optimization in the Space Spanned by the Intensity Functions	37
	1.7.3 Results	38
	1.8 Conclusion	42
	References	44
	2 Investigating Functional Cooperation in the Human Brain Using Simple Graph-Theoretic Methods	47
	2.1 Introduction and Background	47
	2.2 Graph Theory and Neuroscience	49
	2.3 A Database of Imaging Experiments	51
	2.4 The Usefulness of Co-activation Graphs	53
	2.5 Relating fMRI to EEG	55
	2.6 Conclusion	57
	References	57
	3 Methodological Framework for EEG Feature Selection Based on Spectral and Temporal Profiles	59
	3.1 Introduction	60
	3.2 Methods	61
	3.2.1 Methodology Overview	61
	3.2.2 Feature Extraction (Step 1)	62
	3.2.3 Feature Selection (Step 2)	66
	3.2.4 Feature Refinement (Steps 3 and 4)	66
	3.3 Results	68
	3.3.1 Simulation Test	68
	3.4 Discussion	69
	3.5 Conclusion	70
	References	71
	4 Blind Source Separation of Concurrent Disease-Related Patterns from EEG in Creutzfeldt--Jakob Disease for Assisting Early Diagnosis	73
	4.1 Introduction	74
	4.2 Patients and EEG Recordings	78
	4.3 Methods	80
	4.3.1 Independent Component Analysis and Extractionof CJD-Related Components	80
	4.3.2 Bayesian Information Criterion	81
	4.4 Results	82
	4.4.1 Determination of the Number of Sources	82
	4.4.2 CJD-Related Feature Extraction	83
	4.4.3 Feature Extraction by PCA	85
	4.5 Discussions	85
	4.6 Conclusions	88
	References	89
	5 Comparison of Supervised Classification Methods with Various Data Preprocessing Procedures for Activation Detectionin fMRI Data	91
	5.1 Introduction	91
	5.2 Data Set	92
	5.3 Data Preprocessing	93
	5.4 Pattern Recognition Methods	94
	5.4.1 Fisher Linear Discriminant	95
	5.4.2 Support Vector Machine	95
	5.4.3 Gaussian Nave Bayes	96
	5.4.4 Correlation Analysis	96
	5.4.5 k-Nearest Neighbor	96
	5.5 Results	97
	5.6 Conclusions	98
	References	98
	6 Recent Advances of Data Biclustering with Application in Computational Neuroscience	100
	6.1 Introduction	100
	6.1.1 Motivation	100
	6.1.2 Data Input	101
	6.1.3 Objective of Task	102
	6.1.4 History	103
	6.1.5 Outline	104
	6.2 Biclustering Types and Structures	104
	6.2.1 Notations	104
	6.2.2 Bicluster Types	105
	6.2.3 Biclustering Structures	107
	6.3 Biclustering Techniques and Algorithms	109
	6.3.1 Based on Matrix Means and Residues	109
	6.3.2 Based on Matrix Ordering, Reordering, and Decomposition	111
	6.3.3 Based on Bipartite Graphs	115
	6.3.4 Based on Information Theory	118
	6.3.5 Based on Probability	119
	6.3.6 Comparison of Biclustering Algorithms	121
	6.4 Application of Biclustering in Computational Neuroscience	122
	6.5 Conclusions	124
	References	124
	7 A Genetic Classifier Account for the Regulation of Expression	128
	7.1 Introduction	128
	7.1.1 Motivation	128
	7.1.2 Background	129
	7.2 Model and Methods	130
	7.2.1 Basic Assumptions	130
	7.2.2 Model Structure	130
	7.2.3 Model Equations	131
	7.2.4 Stability	132
	7.3 Results	132
	7.3.1 Composition by Overlap of Nodes	132
	7.3.1.1 Complete Overlap	132
	7.3.1.2 Incomplete Overlap	134
	7.3.2 Multiple Gene Scenarios	134
	7.3.2.1 Three Genes	134
	7.3.3 Composition by Infinite Chains	135
	7.3.3.1 Chain of Genes Including A 1-Product Gene	136
	7.3.3.2 Chain of Genes Without A 1-Product Gene	136
	7.3.4 Subchains	137
	7.4 Discussion	137
	References	138
	Part II Modeling	139
	8 Neuroelectromagnetic Source Imaging of Brain Dynamics	140
	8.1 Introduction	140
	8.1.1 Neuronal Origins of Electromagnetic Signals	141
	8.2 Measurement Modalities	142
	8.2.1 Magnetoencephalography (MEG)	142
	8.2.2 Electroencephalography (EEG)	143
	8.2.3 Electrocorticography (ECoG)	143
	8.3 Data Preprocessing	143
	8.4 Overview of Modeling Steps	145
	8.4.1 Modeling of Neural Generators	145
	8.4.2 Anatomical Modeling of Head Tissues and Neural Sources	146
	8.4.3 Multimodal Geometric Registration	146
	8.4.4 Forward Modeling	147
	8.4.5 Inverse Modeling	147
	8.5 Parametric Dipole Modeling	148
	8.5.1 Uncorrelated Noise Model	148
	8.5.2 Correlated Noise Model	149
	8.5.3 Global Minimization	150
	8.6 Source Space-Based Distributed and Sparse Methods	150
	8.6.1 Bayesian Maximum a Posteriori (MAP) Estimates	151
	8.6.2 Dynamic Statistical Parametric Mapping (dSPM)	154
	8.6.3 Standardized Low Resolution Brain Electromagnetic Tomography (sLORETA)	155
	8.6.4 Sparse Bayesian Learning (SBL) and Automatic Relevance Determination (ARD)	156
	8.7 Spatial Scanning and Beamforming	158
	8.7.1 Matched Filter	159
	8.7.2 Multiple Signal Classification (MUSIC)	159
	8.7.3 Linearly Constrained Minimum Variance (LCMV) Beamforming	160
	8.7.4 Synthetic Aperture Magnetometry (SAM)	160
	8.7.5 Dynamic Imaging of Coherent Sources (DICS)	161
	8.7.6 Other Spatial Filtering Methods	161
	8.8 Comparison of Methods	162
	8.9 Conclusion	162
	References	164
	9 Optimization in Brain? -- Modeling Human Behavior and Brain Activation Patterns with Queuing Network and Reinforcement Learning Algorithms	169
	9.1 Introduction	169
	9.2 Modeling Behavioral and Brain Imaging Phenomena in Transcription Typing with Queuing Networks and Reinforcement Learning Algorithms	171
	9.2.1 Behavioral Phenomena	171
	9.2.2 Brain Imaging Phenomena	171
	9.2.3 A Queuing Network Model with Reinforcement Learning Algorithms	172
	9.2.3.1 The Static Portion of the Queuing Network Model	172
	9.2.3.2 The Dynamic Portion of the Queuing Network Model: Self-Organization of the Queuing Network with Reinforcement Learning Algorithms	173
	9.2.4 Model Predictions of three Skill Learning Phenomenaand two Brain Imaging Phenomena	176
	9.2.4.1 Predictions of the three Skill Learning Phenomena	176
	9.2.4.2 Predictions of the First Brain Imaging Phenomenon	177
	9.2.4.3 Predictions of the Second Brain Imaging Phenomenon	177
	9.2.5 Simulation of the three Skill Learning Phenomenaand the two Brain Imaging Phenomena	178
	9.2.5.1 The First and the Second Skill Learning Phenomena	178
	9.2.5.2 The Third Skill Learning Phenomena	178
	9.2.5.3 The First Brain Imaging Phenomena	179
	9.2.5.4 The Second Brain Imaging Phenomena	179
	9.3 Modeling the Basic PRP and Practice Effect on PRP with Queuing Networks and Reinforcement Learning Algorithms	180
	9.3.1 Modeling the Basic PRP and the Practice Effect on PRPwith Queuing Networks	180
	9.3.1.1 Learning Process in Individual Servers	182
	9.3.1.2 Learning Process in the Simplest Queuing Network with two Routes	183
	9.3.2 Predictions of the Basic PRP and the Practice Effect on PRP with the Queuing Network Model	184
	9.3.3 Simulation Results	184
	9.4 Discussion	186
	References	189
	10 Neural Network Modeling of Voluntary Single-Joint Movement Organization I. Normal Conditions	192
	10.1 Introduction	192
	10.2 Models and Theories of Motor Control	193
	10.3 The Extended VITE--FLETE Models Without Dopamine	195
	10.4 Conclusion	200
	References	200
	11 Neural Network Modeling of Voluntary Single-Joint Movement Organization II. Parkinson's Disease	203
	11.1 Introduction	203
	11.2 Brain Anatomy in Parkinson's Disease	204
	11.3 Empirical Signatures	206
	11.4 Is There Dopaminergic Innervation of the Cortexand Spinal Cord?	206
	11.5 Effects of Dopamine Depletion on Neuronal, Electromyographic, and Movement Parameters in PD Humans and MPTP Animals	207
	11.5.1 Cellular Disorganization in Cortex	207
	11.5.2 Reduction of Neuronal Intensity and of Rate of Development of Neuronal Discharge in the PrimaryMotor Cortex	208
	11.5.3 Significant Increase in Mean Duration of Neuronal Discharge in Primary Motor Cortex Preceding and Following Onset of Movement	208
	11.5.4 Prolongation of Behavioral Simple Reaction Time	209
	11.5.5 Repetitive Triphasic Pattern of Muscle Activation	210
	11.5.6 Electromechanical Delay Time Is Increased	210
	11.5.7 Depression of Rate of Development and Peak Amplitudeof the First Agonist Burst of EMG Activity	210
	11.5.8 Movement Time Is Significantly Increased	211
	11.5.9 Reduction of Peak Velocity	212
	11.5.10 Reduction of Peak Force and Rate of Force Production	212
	11.5.11 Movement Variability	212
	11.6 The Extended VITE--FLETE Models with Dopamine	213
	11.7 Simulated Effects of Dopamine Depletion on the Cortical Neural Activities	216
	11.8 Simulated Effects of Dopamine Depletion on EMG Activities	217
	11.9 Conclusion	219
	References	220
	12 Parametric Modeling Analysis of Optical Imaging Data on Neuronal Activities in the Brain	223
	12.1 Introduction	224
	12.2 Methods	226
	12.2.1 Recording of Optical Signals and Preprocessing	226
	12.2.2 Modeling	228
	12.2.3 Classification of Optical Signals Based on Activation Timing	229
	12.3 Results	231
	12.3.1 Estimation of STF Model Parameters	231
	12.3.2 Classification of Pixel Activity Patterns	232
	12.4 Discussion	234
	References	234
	13 Advances Toward Closed-Loop Deep Brain Stimulation	236
	13.1 Introduction	236
	13.2 Nerve Stimulation	237
	13.3 Local Field Potentials	238
	13.4 Parkinson's Disease	239
	13.4.1 Treatments	240
	13.5 Deep Brain Stimulation	241
	13.5.1 DBS Mechanism	241
	13.5.2 Apparatus	241
	13.5.3 Stimulus Specifications	242
	13.5.4 DBS Programming	244
	13.5.5 Side Effects	246
	13.6 Biosignal Processing	246
	13.6.1 Features	247
	13.6.2 Classifiers	247
	13.6.3 Feature Selection	248
	13.7 Closed-Loop DBS	248
	13.7.1 Demand-Controlled DBS	249
	13.7.2 ALOPEX and DBS	250
	13.7.3 Genetic Algorithms and DBS	251
	13.7.4 Hardware Implementations	251
	13.8 Related Advances in Other Neuroprosthetic Research	252
	13.8.1 Closed-Loop Cardiac Pacemaker Technology	253
	13.8.2 Brain-to-Computer Interface	253
	13.9 Neural Network Modeling and the Basal Ganglia	254
	13.10 Summary	255
	References	255
	14 Molecule-Inspired Methods for Coarse-Grain Multi-System Optimization	263
	14.1 Introduction	264
	14.2 Biomolecular Computing In Vitro	265
	14.3 Biomolecular Computing In Silico	266
	14.4 Neural Nets in Biomolecules	268
	14.5 Conclusions and Future Work	272
	References	274
	Part III Brain Dynamics/Synchronization	276
	15 A Robust Estimation of Information Flow in Coupled Nonlinear Systems	277
	15.1 Introduction	277
	15.2 Methodology	279
	15.2.1 Transfer Entropy (TE)	279
	15.2.2 Improved Computation of Transfer Entropy	280
	15.2.2.1 Selection of k	280
	15.2.2.2 Selection of l	280
	15.2.2.3 Selection of Radius r	281
	15.2.3 Statistical Significance of Transfer Entropy	283
	15.2.4 Detecting Causality Using Transfer Entropy	284
	15.3 Simulation Example	284
	15.3.1 Statistical Significance of TE and NTE	285
	15.3.2 Robustness to Noise	287
	15.4 Discussion and Conclusion	288
	References	289
	16 An Optimization Approach for Finding a Spectrum of Lyapunov Exponents	290
	16.1 Introduction	290
	16.2 Lyapunov Exponents	291
	16.3 An Optimization Approach	293
	16.3.1 Theory	294
	16.3.2 Implementation Details	295
	16.3.2.1 Phase Space Reconstruction	296
	16.4 Models Used in the Computational Experiments	299
	16.4.1 Lorenz Attractor	299
	16.4.2 Rössler Attractor	300
	16.4.3 Hénon Map	301
	16.4.4 The Hénon--Heiles Equations	301
	16.5 Computational Experiments	302
	16.5.1 Numerical Computations	302
	16.5.2 Sensitivity Analysis	304
	16.6 Summary and Conclusion	306
	References	306
	17 Dynamical Analysis of the EEG and Treatment of Human Status Epilepticus by Antiepileptic Drugs	309
	17.1 Introduction	310
	17.2 Materials and Methods	311
	17.2.1 Recording Procedure and EEG Data	311
	17.2.1.1 EEG from Barrow Neurological Institute, Phoenix, Arizona	312
	17.2.1.2 EEG from Mayo Clinic Hospital, Scottsdale, Arizona	312
	17.2.2 Measures of Brain Dynamics	313
	17.2.2.1 Measure of Chaos(STLmax)	313
	17.2.2.2 Measure of Dynamical Entrainment	315
	17.3 Results	315
	17.4 Conclusion	317
	References	318
	18 Analysis of Multichannel EEG Recordings Based on Generalized Phase Synchronization and Cointegrated VAR	320
	18.1 Introduction	320
	18.2 Integrated and Cointegrated VAR	322
	18.2.1 Augmented Dickey--Fuller Test for Testing the Null Hypothesis of a Unit Root	323
	18.2.2 Estimation of Cointegrated VAR(p) Processes	325
	18.2.3 Testing for the Rank of Cointegration	327
	18.3 The Role of Phase Synchronization in Neural Dynamics	328
	18.4 Phase Estimation Using Hilbert Transform	329
	18.5 Multivariate Approach to Phase Synchrony via Cointegrated VAR	330
	18.5.1 Cointegration Rank as a Measure of Synchronization among Different EEG Channels	331
	18.5.2 Absence Seizures	334
	18.5.3 Numerical Study of Synchrony in Multichannel EEG Recordings from Patients with Absence Epilepsy	334
	18.6 Conclusion	338
	18.6.1 Phillips--Ouliaris Cointegration Test	339
	References	341
	19 Antiepileptic Therapy Reduces Coupling Strength Among Brain Cortical Regions in Patients with Unverricht--Lundborg Disease: A Pilot Study	343
	19.1 Introduction	344
	19.2 Data Information	347
	19.3 Synchronization Measures	348
	19.3.1 Mutual Information	348
	19.3.2 Nonlinear Interdependencies	350
	19.4 Statistical Tests and Data Analysis	351
	19.5 Conclusion and Discussion	354
	References	355
	20 Seizure Monitoring and Alert System for Brain Monitoring in an Intensive Care Unit	358
	20.1 Introduction	359
	20.2 Preictal Transition and Seizure Prediction	360
	20.3 Methods	362
	20.3.1 Chaos Theory and Epilepsy	362
	20.3.2 Statistical Method for Pairwise Comparison of STLMAX	364
	20.3.3 Finding Critical Sites by Quadratic Optimization Approach	365
	20.4 Two Main Components of the Seizure Monitoring and Alert System	366
	20.4.1 Algorithm for Generating Automatic Warnings about Impending Seizure from EEG	367
	20.5 Conclusions	368
	References	368