Advances in Neural Computation, Machine Learning, and Cognitive Research : Selected Papers From the XIX International Conference on Neuroinformatics, October 2-6, 2017, Moscow, Russia

This book describes new theories and applications of artificial neural networks, with a special focus on neural computation, cognitive science and machine learning. It discusses cutting-edge research at the intersection between different fields, from topics such as cognition and behavior, motivation and emotions, to neurocomputing, deep learning, classification and clustering. Further topics include signal processing methods, robotics and neurobionics, and computer vision alike. The book includes selected papers from the XIX International Conference on Neuroinformatics, held on October 2-6, 2017, in Moscow, Russia. Preface 6 Organization 7 Editorial Board 7 Advisory Board 7 Program Committee of the XIX International Conference “Neuroinformatics-2017” 9 General Chair 9 Co-chairs 9 Program Committee 9 Contents 12 Neural Network Theory 15 The Analysis of Regularization in Deep Neural Networks Using Metagraph Approach 16 Abstract 16 1 Introduction 16 2 Description of the Used Neural Network 17 3 Regularization of Deep Neural Networks 17 4 Regularization Representation Using Metagraph Approach 18 5 Experiments 20 6 Conclusion 21 References 21 Adding Noise During Training as a Method to Increase Resilience of Neural Network Solution of Invers ... 22 Abstract 22 1 Introduction 22 2 The Initial Data and the Statement of the IP of MTS 24 3 Description of the Noise 24 4 The Use of Artificial Neural Networks 25 5 Results 25 6 Conclusion 27 Acknowledgement 28 References 28 Multi-Layer Solution of Heat Equation 30 1 Introduction 30 2 Model Problem 32 3 Calculations 33 4 Conclusions 34 References 34 Implementation of a Gate Neural Network Based on Combinatorial Logic Elements 36 Abstract 36 1 Introduction 36 2 Model of the Gate Neural Network 37 3 Network Learning Algorithm 41 4 Analysis of the Results 42 5 Conclusion 43 References 44 Adaptive Gateway Element Based on a Recurrent Neurodynamical Model 46 1 Introduction 46 2 Neuron Model 47 3 Gateway Model 49 4 Conclusions 51 References 51 Common Sense Knowledge in Large Scale Neural Conversational Models 52 Abstract 52 1 Introduction 52 2 Methods and Algorithms 53 2.1 Datasets 53 2.2 Neural Network Architectures 53 2.3 Word Vectors 54 3 Results and Discussion 54 3.1 Reply Selection Accuracies 54 3.2 Factoid Answer Selection from Alternatives 54 3.3 Common Sense Questions 55 4 Conclusions 56 References 57 Applications of Neural Networks 58 Prospects for the Development of Neuromorphic Systems 59 Abstract 59 1 Introduction 59 2 Overview of Deep Neural Network Architectures 60 3 Neuron Models 61 4 Discussion 62 References 62 Pulse Neuron Learning Rules for Processing of Dynamical Variables Encoded by Pulse Trains 65 Abstract 65 1 Introduction 65 2 Problem Formulation 66 3 Temporal Learning Rules of the Pulse Neuron 67 4 Computer Simulation 67 5 Conclusions 69 References 70 Information Environment for Neural-Network Adaptive Control System 71 Abstract 71 1 Introduction 71 2 Analysis of the Structure of a Digital Neural Network Control System Based on the Universal Computer 73 3 Implementation of Phases of the Control Cycle of the Neural Network System 74 4 A Coherent Information Environment Model for Neural-Network Control System 75 5 Conclusion 76 References 76 Neural Network Semi-empirical Modeling of the Longitudinal Motion for Maneuverable Aircraft and Identification of Its Aerodynamic Characteristics 77 1 Introduction 77 2 Mathematical Model of Longitudinal Motion for Maneuverable Aircraft 78 3 Generation of a Representative Set of Training Data 79 4 Semi-empirical Neural Network Model of Aircraft Longitudinal Motion 80 5 Conclusions 82 References 82 Dump Truck Fault's Short-Term Forecasting Based on the Multi-agent Adaptive Fuzzy Neuronet 84 1 Introduction 84 2 The Multi-agent Adaptive Fuzzy Neuronet for Dump Truck Fault's Short-Term Forecasts 85 2.1 The Training Algorithms of the Multi-agent Adaptive Fuzzy Neuronet 86 2.2 The Multi-agent Adaptive Fuzzy Neuronet 88 3 Results 89 References 90 Object Detection on Images in Docking Tasks Using Deep Neural Networks 91 Abstract 91 1 Introduction 91 1.1 Relevance of the Problem 91 1.2 Statement of the Problem 92 2 Description of Chosen Systems 93 2.1 Structure of the Faster R-CNN 93 2.2 Learning of Faster R-CNN 94 3 Experimental Researches 94 4 Conclusions 95 References 96 Detection of Neurons on Images of the Histological Slices Using Convolutional Neural Network 97 Abstract 97 1 Introduction 97 2 Methods 99 3 Results 99 4 Conclusions 101 Acknowledgements 101 References 101 Constructing a Neural-Net Model of Network Traffic Using the Topologic Analysis of Its Time Series Complexity 103 Abstract 103 1 Introduction 103 2 The Topological Data Analysis 104 3 Setting the Problem 105 4 Topological Invariants Calculated for a Traffic Intensity Sequence 106 5 Building the Neural-Net Model of the Data 108 6 Conclusions 109 References 109 Texture Recognition from Positions of the Theory of Active Perceptions 110 1 Introduction 110 2 Using TAP in Image Recognition 111 3 Formation of Feature Description of a Textured Image 112 4 Computational Experiment 113 5 Conclusion 115 References 115 Method of Calculation of Upper Bound of Learning Rate for Neural Tuner to Control DC Drive 116 Abstract 116 1 Introduction 116 2 Definition of the Tuner and Problem Statement 117 3 Stability of the Control System 117 4 Upper Bound of Learning Rate Calculation 118 5 Experimental Results 119 6 Conclusion 121 Acknowledgments 121 References 121 Intelligent Diagnostics of Mechatronic System Components of Career Excavators in Operation 122 Abstract 122 1 Introduction 122 2 The Organization of the Intelligent Diagnostics of Mechatronic Complex Components 123 3 Neural Network for Data Processing 124 4 Practical Implementation of Algorithms in the Diagnostic System 127 5 Conclusion 127 References 128 Emotion Recognition in Sound 129 Abstract 129 1 Introduction 129 2 Materials and Methods 129 3 Examined Approach 132 4 Conclusions and Directions for Further Work 135 Acknowledgments 135 References 135 The Classification of Objects Based on a Model of Perception 137 Abstract 137 1 Introduction 137 2 Theoretical Problems 138 2.1 Statement of the Problem 138 2.2 Initial Stages of Information Transformation 138 2.3 Compression of Information by Granulation 139 2.4 Classification of Objects as a Prototype Search 140 3 Properties of the Model 141 4 Conclusion 142 References 142 An Approach to Use Convolutional Neural Network Features in Eye-Brain-Computer-Interface 144 Abstract 144 1 Introduction 144 2 Problem Statement 145 3 Representation of the EEG Signals as Images 145 4 Quality of Features Generated by the Convolutional Neural Network 147 5 Conclusions 148 Acknowledgements 149 References 149 Semi-empirical Neural Network Model of Real Thread Sagging 150 Abstract 150 1 Introduction 150 2 Semi-empirical Model of a Sagging Thread. Methods 151 3 Calculation 153 4 Conclusions 154 References 155 Cognitive Sciences and Adaptive Behavior 157 Color or Luminance Contrast -- What Is More Important for Vision? 158 1 Introduction 158 2 Contrast of a Color Image 159 3 Color Analog of Rayleigh Criterion 165 4 Conclusion 166 References 166 Synchrony of Cortical Alpha and Beta Oscillations 168 Abstract 168 1 Introduction 168 2 Materials and Methods 169 3 Results 169 4 Conclusions 172 Acknowledgments 172 References 172 Processes of Self-organization in the Community of Investors and Producers 174 Abstract 174 1 Introduction 174 2 Description of the Model 175 2.1 General Scheme of the Model 175 2.2 Description of the Iterative Process 176 3 Results of Computer Simulation 177 4 Conclusion 179 Acknowledgments 180 References 180 Neurobiology 181 Complexity of Heart Rate During More and Less Differentiated Behaviors 182 Abstract 182 1 Introduction 182 2 Materials and Methods 184 3 Results 185 4 Conclusion 186 Acknowledgments 187 References 187 Comparison of Some Fractal Analysis Methods for Studying the Spontaneous Activity in Medullar Audito ... 189 Abstract 189 1 Introduction 189 2 Methods 190 2.1 Electrophysiological Recording of the Background Firing 190 2.2 Data Processing 190 3 Conclusion 193 References 194 Synapse as a Multi-component and Multi-level Information System 195 Abstract 195 1 Introduction 195 2 The Own Goals of the Individual Neuron 196 3 The Functional Systems of the Neuron Involved in Synaptic Modulations in the Early Phase of LTP 197 4 Conclusion 199 Acknowledgements 199 References 200 Effect of Persistent Sodium Current on Neuronal Activity 202 Abstract 202 1 Introduction 202 2 Methods 203 2.1 Experiment 203 2.2 Analysis of Experimental Data 203 2.3 Hodgkin-Huxley-like Model of a Neuron 203 3 Results 204 3.1 Dynamic-Clamp Study of the Influence of NaP Current 204 3.2 Effect of Persistent-Sodium Current in a Modeled Neuron 205 4 Conclusion 207 Acknowledgments 207 References 207 Preface......Page 6 Advisory Board......Page 7 Program Committee......Page 9 Contents......Page 12 Neural Network Theory......Page 15 1 Introduction......Page 16 3 Regularization of Deep Neural Networks......Page 17 4 Regularization Representation Using Metagraph Approach......Page 18 5 Experiments......Page 20 References......Page 21 1 Introduction......Page 22 3 Description of the Noise......Page 24 5 Results......Page 25 6 Conclusion......Page 27 References......Page 28 1 Introduction......Page 30 2 Model Problem......Page 32 3 Calculations......Page 33 References......Page 34 1 Introduction......Page 36 2 Model of the Gate Neural Network......Page 37 3 Network Learning Algorithm......Page 41 4 Analysis of the Results......Page 42 5 Conclusion......Page 43 References......Page 44 1 Introduction......Page 46 2 Neuron Model......Page 47 3 Gateway Model......Page 49 References......Page 51 1 Introduction......Page 52 2.2 Neural Network Architectures......Page 53 3.2 Factoid Answer Selection from Alternatives......Page 54 3.3 Common Sense Questions......Page 55 4 Conclusions......Page 56 References......Page 57 Applications of Neural Networks......Page 58 1 Introduction......Page 59 2 Overview of Deep Neural Network Architectures......Page 60 3 Neuron Models......Page 61 References......Page 62 1 Introduction......Page 65 2 Problem Formulation......Page 66 4 Computer Simulation......Page 67 5 Conclusions......Page 69 References......Page 70 1 Introduction......Page 71 2 Analysis of the Structure of a Digital Neural Network Control System Based on the Universal Computer......Page 73 3 Implementation of Phases of the Control Cycle of the Neural Network System......Page 74 4 A Coherent Information Environment Model for Neural-Network Control System......Page 75 References......Page 76 1 Introduction......Page 77 2 Mathematical Model of Longitudinal Motion for Maneuverable Aircraft......Page 78 3 Generation of a Representative Set of Training Data......Page 79 4 Semi-empirical Neural Network Model of Aircraft Longitudinal Motion......Page 80 References......Page 82 1 Introduction......Page 84 2 The Multi-agent Adaptive Fuzzy Neuronet for Dump Truck Fault's Short-Term Forecasts......Page 85 2.1 The Training Algorithms of the Multi-agent Adaptive Fuzzy Neuronet......Page 86 2.2 The Multi-agent Adaptive Fuzzy Neuronet......Page 88 3 Results......Page 89 References......Page 90 1.1 Relevance of the Problem......Page 91 1.2 Statement of the Problem......Page 92 2.1 Structure of the Faster R-CNN......Page 93 3 Experimental Researches......Page 94 4 Conclusions......Page 95 References......Page 96 1 Introduction......Page 97 3 Results......Page 99 References......Page 101 1 Introduction......Page 103 2 The Topological Data Analysis......Page 104 3 Setting the Problem......Page 105 4 Topological Invariants Calculated for a Traffic Intensity Sequence......Page 106 5 Building the Neural-Net Model of the Data......Page 108 References......Page 109 1 Introduction......Page 110 2 Using TAP in Image Recognition......Page 111 3 Formation of Feature Description of a Textured Image......Page 112 4 Computational Experiment......Page 113 References......Page 115 1 Introduction......Page 116 3 Stability of the Control System......Page 117 4 Upper Bound of Learning Rate Calculation......Page 118 5 Experimental Results......Page 119 References......Page 121 1 Introduction......Page 122 2 The Organization of the Intelligent Diagnostics of Mechatronic Complex Components......Page 123 3 Neural Network for Data Processing......Page 124 5 Conclusion......Page 127 References......Page 128 2 Materials and Methods......Page 129 3 Examined Approach......Page 132 References......Page 135 1 Introduction......Page 137 2.2 Initial Stages of Information Transformation......Page 138 2.3 Compression of Information by Granulation......Page 139 2.4 Classification of Objects as a Prototype Search......Page 140 3 Properties of the Model......Page 141 References......Page 142 1 Introduction......Page 144 3 Representation of the EEG Signals as Images......Page 145 4 Quality of Features Generated by the Convolutional Neural Network......Page 147 5 Conclusions......Page 148 References......Page 149 1 Introduction......Page 150 2 Semi-empirical Model of a Sagging Thread. Methods......Page 151 3 Calculation......Page 153 4 Conclusions......Page 154 References......Page 155 Cognitive Sciences and Adaptive Behavior......Page 157 1 Introduction......Page 158 2 Contrast of a Color Image......Page 159 3 Color Analog of Rayleigh Criterion......Page 165 References......Page 166 1 Introduction......Page 168 3 Results......Page 169 References......Page 172 1 Introduction......Page 174 2.1 General Scheme of the Model......Page 175 2.2 Description of the Iterative Process......Page 176 3 Results of Computer Simulation......Page 177 4 Conclusion......Page 179 References......Page 180 Neurobiology......Page 181 1 Introduction......Page 182 2 Materials and Methods......Page 184 3 Results......Page 185 4 Conclusion......Page 186 References......Page 187 1 Introduction......Page 189 2.2 Data Processing......Page 190 3 Conclusion......Page 193 References......Page 194 1 Introduction......Page 195 2 The Own Goals of the Individual Neuron......Page 196 3 The Functional Systems of the Neuron Involved in Synaptic Modulations in the Early Phase of LTP......Page 197 Acknowledgements......Page 199 References......Page 200 1 Introduction......Page 202 2.3 Hodgkin-Huxley-like Model of a Neuron......Page 203 3.1 Dynamic-Clamp Study of the Influence of NaP Current......Page 204 3.2 Effect of Persistent-Sodium Current in a Modeled Neuron......Page 205 References......Page 207