Time-Space, Spiking Neural Networks and Brain-inspired Artificial Intelligence

Foreword......Page 3
Preface......Page 4
About the Book Content by Topics and Chapters and The Pathway of Knowledge......Page 8
Contents......Page 9
About the Author......Page 0
Time-Space and AI. Artificial Neural Networks......Page 24
1.1 Evolving Processes in Time-Space......Page 25
1.1.1 What Are Evolving Processes?......Page 26
1.1.2 Evolving Processes in Living Organisms......Page 27
1.1.3 Spatio-temporal and Spectro-temporal Evolving Processes......Page 30
1.2 Characteristics of Evolving Processes: Frequency, Energy, Probability, Entropy and Information......Page 31
1.3 Light and Sound......Page 37
1.4 Evolving Processes in Time-Space and Direction......Page 40
1.5 From Data and Information to Knowledge......Page 41
1.6.1 Defining Deep Knowledge in Time-Space......Page 44
1.6.2 How Deep?......Page 47
1.7 Statistical, Computational Modelling of Evolving Processes......Page 48
1.7.1 Statistical Methods for Computational Modelling......Page 49
1.7.2 Global, Local and Transductive (“Personalised”) Modelling [28]......Page 50
1.7.3 Model Validation......Page 53
1.8 Brain-Inspired AI......Page 54
Acknowledgements......Page 57
References......Page 58
2.1 Classical Artificial Neural Networks: SOM, MLP, CNN, RNN......Page 60
2.1.1 Unsupervised Learning in Neural Networks. Self-organising Maps (SOM)......Page 61
2.1.2 Supervised Learning in ANN. Multilayer Perceptron and the Back Propagation Algorithm......Page 65
2.1.3 Convolutional Neural Networks (CNN)......Page 69
2.1.4 Recurrent and LSTM ANN......Page 70
2.2 Hybrid and Knowledge-Based ANN......Page 71
2.3.1 Principles of ECOS......Page 73
2.3.2 Evolving Self-organising Maps......Page 74
2.3.3 Evolving MLP......Page 77
2.4 Evolving Fuzzy Neural Networks. EFuNN......Page 81
2.5 Dynamic Evolving Neuro-fuzzy Inference Systems—DENFIS......Page 91
2.6 Other ECOS Methods and Systems......Page 96
2.7 Chapter Summary and Further Readings for Deeper Knowledge......Page 98
References......Page 99
The Human Brain......Page 105
3.1 Time-Space in the Brain......Page 106
3.2 Learning and Memory......Page 112
3.3 Neural Representation of Information......Page 114
3.4 Perception in the Brain Is Always Spatio/Spectro-temporal......Page 116
3.5 Deep Learning and Deep Knowledge Representation in Time-Space in the Brain......Page 122
3.6.1 Information Coding......Page 126
3.6.2 Molecular Basis of Information Processing......Page 128
3.7.1 General Notions......Page 130
3.7.2 Electroencephalogram (EEG) Data......Page 132
3.7.4 CT and PET......Page 135
3.7.5 fMRI......Page 136
3.8 Chapter Summary and Further Readings for Deeper Knowledge......Page 138
References......Page 139
Spiking Neural Networks......Page 143
4.1.1 Rate Versus Spike Time Information Representation......Page 144
4.1.2 Spike Encoding Algorithms......Page 146
4.2.1 Hodgkin-Huxley Model (HHM)......Page 154
4.2.2 Leaky Integrate-and-Fire Model (LIFM)......Page 155
4.2.4 Spike Response Model (SRM)......Page 157
4.2.6 Probabilistic and Stochastic Spiking Neuron Models......Page 159
4.2.7 Probabilistic Neurogenetic Model of a Neuron......Page 160
4.3 Methods for Learning in SNN......Page 162
4.3.1 SpikeProp......Page 163
4.3.2 Spike-Time Dependent Plasticity (STDP)......Page 164
4.3.4 Rank Order (RO) Learning Rule......Page 166
4.3.5 Learning in Dynamic Synapses......Page 167
4.4.1 Principles of Spike Pattern Association Learning. The SPAN Model......Page 168
4.4.2 Case Study Examples......Page 172
4.4.3 Memory Capacity of SPAN......Page 175
4.4.4 SPAN for Classification Problems......Page 177
4.5 Why Use SNN?......Page 179
4.6 Summary and Further Readings for Deeper Knowledge......Page 180
References......Page 181
5.1 Principles and Methods of Evolving SNN (ESNN)......Page 185
5.2 Convolutional ESNN (CeSNN)......Page 191
5.3 Dynamic Evolving SNN (DeSNN)......Page 195
5.4.1 Fuzzy Rule Extraction from ESNN......Page 199
5.4.2 A Case Study of Fuzzy Rule Extraction from Water Tastant Sensory Data......Page 204
5.5.1 Reservoir Architectures. Liquid State Machines (LSM)......Page 209
5.5.2 ESNN/DeSNN as Classification/Regression Systems for Reservoir Architectures......Page 211
5.6 Chapter Summary and Further Readings for Deeper Knowledge......Page 213
References......Page 214
6.1.1 A General Architecture of a BI-SNN......Page 216
6.1.2 The BI-SNN NeuCube as a Generic Spatio-temporal Data Machine......Page 218
6.1.3 Mapping Input Temporal Variables into a 3D SNNcube Based on Graph Matching Optimisation Algorithm......Page 226
6.2.1 Deep Unsupervised Learning in Time-Space and Deep Knowledge Representation from Temporal or Spatio/Spectro Temporal Data (TSTD)......Page 232
6.2.2 Deep Supervised Learning in Time-Space......Page 235
6.2.3 Deep Learning in Time-Space for Predictive Modelling in NeuCube. The EPUSSS Algorithm......Page 236
6.3.1 Event-Based Modelling. External Versus Internal Time. Past-, Present- and Future Time......Page 241
6.4 A Design Methodology for Application Oriented Spatio-temporal Data Machines......Page 242
6.4.1 Design Methodology for Implementing Application Oriented Spatio-temporal Data Machines as BI-AI Systems in NeuCube......Page 244
6.4.2 Input Data Encoding......Page 245
6.4.3 Spatial Mapping of Input Variables......Page 247
6.4.5 Supervised Training and Classification/Regression of Dynamic Spiking Patterns of the SNNcube in a SNN Classifier......Page 248
6.4.6 3D Visualisation of the SNNcube......Page 249
6.4.7 Optimisation of NeuCube Structure and Parameters......Page 250
6.5 Case Studies of the Design and Implementation of Classification and Regression Spatio-temporal Data Machines......Page 251
6.5.2 A Case Study on the Design a Regression/Prediction Spatio-temporal Data Machine in NeuCube......Page 252
6.6 Chapter Summary and Further Readings for Deeper Knowledge......Page 253
Acknowledgements......Page 256
References......Page 257
7 Evolutionary- and Quantum-Inspired Computation. Applications for SNN Optimisation......Page 259
7.1.1 The Origin and the Evolution of Life......Page 260
7.1.2 Methods of Evolutionary Computation (EC)......Page 261
7.1.3 Genetic Algorithms......Page 263
7.1.4 Evolutionary Strategies (ES)......Page 265
7.1.5 Particle Swarm Optimisation......Page 266
7.1.6 Estimation of Distribution Algorithms (EDA)......Page 268
7.1.7 Artificial Life Systems......Page 269
7.2.1 Principles of Quantum Information Processing......Page 270
7.2.3 Quantum Inspired Evolutionary Algorithm (QiEA)......Page 273
7.2.4 Versatile QiEA (VQiEA)......Page 276
7.2.5 Extension of the VQiEA to Deal with Continuous Value Variables......Page 279
7.3.1 A Quantum-Inspired Representation of a SNN......Page 282
7.3.2 Application of QiEA for the Optimisation of ESNN Classifier on Ecological Data......Page 285
7.3.3 Integrative Computational Neuro Genetic Model (CNGM) Utilising Quantum-Inspired Representation......Page 286
7.4.1 Quantum Inspired Particle Swarm Optimisation Algorithms......Page 288
7.4.2 Quantum Inspired Particle Swarm Optimisation Algorithm (QiPSO) for the Optimisation of ESNN......Page 289
7.4.3 Dynamic QiPSO......Page 291
7.4.4 Application of DQiPSO for Feature Selection and Model Optimisation......Page 292
7.5 Chapter Summary and Further Readings for Deeper Knowledge......Page 296
References......Page 299
Deep Learning and Deep Knowledge Representation of Brain Data......Page 302
8.1.1 Spatio-temporal Brain Data......Page 303
8.1.2 Brain Atlases......Page 304
8.1.3 EEG Data......Page 307
8.2 Deep Learning and Deep Knowledge Representation of EEG Data in BI-SNN......Page 312
8.3.1 System Design......Page 318
8.3.3 Experimental Results......Page 321
8.3.4 Model Interpretation......Page 323
8.4.1 General Notions......Page 324
8.4.2 Using a NeuCube Model for Emotion Recognition......Page 325
8.4.4 Analysis of the Connectivity in a Trained SNNcube When a Person Is Perceiving Emotional Face and When a Person Is Expressing Such Emotions......Page 326
8.5 Deep Learning and Modelling of Peri-perceptual Processes in BI-SNN......Page 329
8.5.1 The Psychology of Sub-conscious Brain Processes......Page 330
8.5.2 Experimental Setting and EEG Data Collection......Page 331
8.5.3 The Design of a NeuCube Model......Page 333
8.6.3 Results......Page 340
Acknowledgements......Page 342
Appendix......Page 343
References......Page 345
9.1 SNN for Modelling EEG Data to Assess a Potential Progression from MCI to AD......Page 350
9.1.2 Design of a NeuCube Model......Page 351
9.1.3 Classification Results......Page 354
9.2 SNN for Predictive Modelling of Response to Treatment Using EEG Data......Page 355
9.2.2 The Case Study Problem Specification and Data Collection......Page 356
9.2.3 Modelling the EEG Data in a NeuCube Model......Page 359
9.2.4 Comparative Analysis of Brain Activities of MMT Subjects Under Different Drug Doses Versus CO and OP Subjects. Modelling and Understanding the Information Exchange Between Brain Areas Measured Through EEG Channels......Page 363
9.2.5 Analysis of Classification Results......Page 366
9.3 Chapter Summary and Further Readings for Deeper Knowledge......Page 367
References......Page 368
10.1.1 What Are fMRI Data?......Page 371
10.1.2 Traditional Methods for fMRI Data Analysis......Page 373
10.1.3 Selecting Features from FMRI Data......Page 375
10.2.1 Why Use SNN for Modelling of fMRI Spatio-temporal Brain Data?......Page 376
10.2.2 A Methodology for Deep Learning and Deep Knowledge Representation of fMRI Data in BI-SNN......Page 377
10.3.1 The STAR/PLUS Benchmark fMRI Data......Page 380
10.3.2 fMRI Data Encoding, Mapping and Learning in a NeuCube SNN Model......Page 381
10.3.3 Classification of the fMRI Data in a NeuCube-Based Model......Page 387
10.4 Algorithms for Modelling fMRI Data that Measure Cognitive Processes......Page 389
10.4.2 Connectivity Initialization and Deep Learning in a SNN Cube......Page 390
10.4.4 A Case Study Implementation on the STAR/PLUS Data......Page 393
10.5 Chapter Summary and Further Readings for Deeper Knowledge......Page 398
Acknowledgements......Page 400
References......Page 401
11.1 Introduction and Background Work......Page 406
11.2 A Personalised Modelling Architecture for fMRI and DTI Data Integration Based on the NeuCube BI-SNN......Page 409
11.3 Orientation-Influence Driven STDP (oiSTDP) Learning in SNN for the Integration of Time-Space and Direction, Illustrated on fMRI and DTI Data......Page 411
11.3.2 Neuron Model......Page 412
11.3.3 Unsupervised Weight Adaptation of Synapses......Page 415
11.4.2 Experimental Results......Page 421
11.5.1 Problem Specification and Data Preparation......Page 423
11.5.2 Modelling and Experimental Results......Page 426
11.6 Chapter Summary and Further Readings for Deeper Knowledge......Page 429
References......Page 430
SNN for Audio-Visual Data and Brain-Computer Interfaces......Page 437
12.1 Audio and Visual Information Processing in the Human Brain......Page 438
12.1.1 Audio Information Processing......Page 439
12.1.2 Visual Information Processing......Page 441
12.1.3 Integrated Audio and Visual Information Processing......Page 444
12.2.1 Issues with Modelling Audio-Visual Information with SNN......Page 447
12.2.2 Convolutional eSNN (CeSNN) for Modelling Visual Information......Page 449
12.2.3 Convolutional eSNN (CeSNN) for Modelling Audio Information......Page 450
12.2.4 Convolutional eSNN (CeSNN) for Integrated Audio-Visual Information Processing......Page 451
12.3.1 Data Sets......Page 455
12.3.2 Experimental Results......Page 456
12.4 Chapter Summary and Further Readings for Deeper Knowledge......Page 460
References......Page 462
13.1.1 Deep Learning of Audio Data in the Brain......Page 464
13.1.3 Deep Learning and Recognition of Music......Page 466
13.1.4 Experimental Results......Page 467
13.2.1 Two Approaches to Visual Information Processing......Page 469
13.2.2 Applications for Fast Moving Object Recognition......Page 470
13.2.3 Applications for Gender and Age Group Classification Based on Face Recognition......Page 471
13.3.2 The Brain-Inspired SNN and the Proposed Retinotopic Mapping......Page 474
13.3.3 Unsupervised and Supervised Learning of Dynamic Visual Patterns......Page 476
13.3.4 Design of an Experiment for the MNIST-DVS Benchmark Dataset......Page 477
13.3.5 Experimental Results......Page 478
13.3.6 Model Interpretation for a Better Understanding of the Processes Inside the Visual Cortex......Page 479
13.4 Chapter Summary and Further Readings for Deeper Knowledge......Page 480
References......Page 481
14.1.1 General Notions......Page 485
14.1.3 Types and Applications of BCI......Page 487
14.2.1 The NeuCube BI-SNN Architecture......Page 491
14.2.2 A Brain-Inspired Framework for BCI (BI-BCI) with Neurofeedback......Page 494
14.3.1 Introduction......Page 495
14.3.2 Design of an Experimental BI-BCI System......Page 497
14.3.4 Analysis of the Results......Page 498
14.4.1 General Notions......Page 500
14.4.2 Applications......Page 502
14.5 From BI-BCI to Knowledge Transfer Between Humans and Machines......Page 504
References......Page 505
SNN in Bio- and Neuroinformatics......Page 509
15.1.1 General Notions......Page 510
15.1.2 DNA, RNA and Proteins. The Central Dogma of Molecular Biology and the Evolution of Life.......Page 511
15.1.3 Phylogenetics......Page 517
15.1.4 The Challenges of Molecular Data Analysis......Page 518
15.2.1 Biological Databases......Page 521
15.2.2 General Information About Bioinformatics Data Modelling......Page 522
15.2.3 Gene Expression Data Modelling and Profiling......Page 524
15.2.4 Clustering of Time Series Gene Expression Data......Page 526
15.2.5 Protein Data Modelling and Structure Prediction......Page 528
15.3.1 General Notions......Page 529
15.3.2 Gene Regulatory Network Modelling......Page 531
15.3.3 Protein Interaction Networks......Page 532
15.4.1 General Notions......Page 534
15.4.2 A SNN Based Methodology for Gene Expression Time Series Data Modelling and Extracting GRN......Page 535
15.4.3 Extracting GRN from a Trained Model......Page 537
15.4.4 A Case Study Experimental Modelling of Gene Expression Time Series Data......Page 538
15.4.5 Extracting GRN Form a Trained Model and Analysis of the GRN for New Knowledge Discovery......Page 540
15.4.6 Discussions on the Method......Page 543
15.5 Chapter Summary and Further Readings......Page 544
References......Page 545
16.1.1 General Notions......Page 549
16.2.1 The PNGM of a Spiking Neuron......Page 552
16.2.2 Using the PNGM of a Neuron to Build SNN......Page 555
16.3.1 CNGM Architectures......Page 556
16.3.2 The NeuCube Architecture as a CNGM......Page 557
16.4.1 Modelling Brain Diseases......Page 559
16.4.2 CNGM for Cognitive Robotics and Emotional Computing......Page 560
16.5 Life, Death and CNGM......Page 561
16.6 Chapter Summary and Further Readings for Deeper Knowledge......Page 562
References......Page 563
17.1.1 Introduction: Global, Local and Personalise Modelling......Page 566
17.1.2 A Framework for Personalised Modelling (PM) Based on Integrated Feature and Model Parameter Optimisation......Page 568
17.2.2 Classification Accuracy and Comparative Analysis......Page 576
17.3.1 Introduction......Page 579
17.3.2 Using SNN and ESNN for PM......Page 581
17.3.3 An ESNN Method for PM on Biomedical Data......Page 583
17.3.4 A Case Study of PM for Chronic Kidney Disease Data Classification......Page 588
Acknowledgements......Page 590
References......Page 591
18.1.1 Introduction......Page 595
18.1.2 A NeuCube-Based Framework for PM of Integrated Static and Dynamic Data......Page 597
18.1.3 Comparative Analysis of the NeuCube Based Method with Other Methods for PM......Page 600
18.2.1 The Case Study Data for Individual Stroke Risk Prediction......Page 601
18.2.2 Personalised Deep Learning and Knowledge Representation in NeuCube on the Case of Stroke......Page 603
18.3.1 The Case Study Problem and Data......Page 606
18.3.2 The NeuCube Based PM Model......Page 607
18.3.3 Experimental Results......Page 608
18.3.4 Discussions......Page 609
Acknowledgements......Page 611
References......Page 612
 Deep in Time-Space Learning and Deep Knowledge Representation of Multisensory Streaming Data......Page 618
19.1 A General Framework for Deep Learning and Predictive Modelling of Multisensory Time-Space Streaming Data with SNN......Page 619
19.1.1 The Challenges of Pattern Recognition and Modelling of Multisensory Streaming Data......Page 620
19.1.2 Modelling Streaming Data in Evolving SNN (eSNN)......Page 621
19.1.3 A General Methodology for Modelling Multisensory Streaming Data in Brain-Inspired SNN for Classification and Regression......Page 622
19.2 Stock Market Movement Prediction Using On-Line Predictive Modelling with eSNN......Page 628
19.3.1 Early Event Prediction in Ecology: General Notions......Page 631
19.3.2 A Case Study on Predicting Abundance of Fruit Flies Using Spatio-temporal Climate Data......Page 632
19.4.2 NeuCube Model Creation and Modelling Results......Page 638
19.5.2 Predictive Modelling of Seismic Data for Earthquake Forecasting Using NeuCube......Page 642
19.5.3 Experiment Design......Page 644
19.5.4 Discussions......Page 648
19.6.1 Modelling Multisensory Air Pollution Streaming Data......Page 649
19.7 Chapter Summary and Further Readings for Deeper Knowledge......Page 651
Acknowledgements......Page 653
Appendix 2......Page 654
References......Page 655
Future Development in BI-SNN and BI-AI......Page 659
20.1.1 General Notions......Page 660
20.1.2 The von Neumann Computation Principle and the Atanassov’s ABC Machine......Page 661
20.2.1 General Principles......Page 663
20.2.2 Hardware Platforms for Neuromorphic Computation......Page 664
20.3.1 A Brief Overview of SNN Development Systems......Page 666
20.3.2 The NeuCube Development System for Spatio-temporal Data Machines......Page 668
20.3.3 Implementation of NeuCube-Based Spatio-temporal Data Machines on Traditional and on Neuromorphic Hardware Platforms......Page 671
20.4 Chapter Summary and Further Readings......Page 672
References......Page 673
21.1 Claude Shannon’s Classical Information Theory......Page 677
21.2 The Proposed Information Theory for Temporal Data Compression for Classification Tasks Based on Spike-Time Encoding......Page 679
21.3 A Spike-Time Encoding and Compression Method for fMRI Spatio-Temporal Data Classification......Page 683
21.4 Chapter Summary and Further Readings......Page 693
Appendix......Page 694
References......Page 695
22.1 Towards Integrated Quantum-Molecular-Neurogenetic-Brain-Inspired Models......Page 698
22.1.1 Quantum Computation......Page 699
22.1.2 The Concept of an Integrated Quantum-Neurogenetic-Brain-Inspired Model Based on SNN......Page 701
22.2.2 Towards a Symbiosis Between Human Intelligence and Artificial Intelligence (HI + AI), Led by HI......Page 704
22.3 Summary and Further Readings for a Deeper Knowledge......Page 708
Epilogue......Page 712
Glossary......Page 714
Index......Page 731