Neural Information Processing : 29th International Conference, ICONIP 2022, Virtual Event, November 22–26, 2022, Proceedings, Part II

The three-volume set LNCS 13623, 13624, and 13625 constitutes the refereed proceedings of the 29th International Conference on Neural Information Processing, ICONIP 2022, held as a virtual event, November 22–26, 2022. The 146 papers presented in the proceedings set were carefully reviewed and selected from 810 submissions. They were organized in topical sections as follows: Theory and Algorithms; Cognitive Neurosciences; Human Centered Computing; and Applications. The ICONIP conference aims to provide a leading international forum for researchers, scientists, and industry professionals who are working in neuroscience, neural networks, deep learning, and related fields to share their new ideas, progress, and achievements. Preface Organization Contents – Part II Cognitive Neurosciences Differences in Brain Activation During Physics Problem Solving Across Students with Various Learning .26em plus .1em minus .1emProgression: Electrophysiological Evidence Based on Detrended Fluctuation Analysis 1 Introduction 2 Methods 2.1 Participants 2.2 Stimuli Materials and Procedure 2.3 EEG Data Acquisition and Data Preprocess 2.4 Detrended Fluctuation Analysis (DFA) and Scaling Exponents 2.5 Data Analysis 3 Results and Discussions 3.1 Behavior Results 3.2 Short-Range Temporal Correlations (SRTCs: 1) 3.3 Long-Range Temporal Correlations (LRTCs: 2) 3.4 Correlation Between DFA Exponents and Accuracy 4 Conclusions References A Dynamic, Economical, and Robust Coding Scheme in the Lateral Prefrontal Neurons of Monkeys 1 Introduction 2 Methods 2.1 Subjects 2.2 Behavioral Task and Neuronal Recording 2.3 Data Analysis 3 Results 3.1 Database 3.2 Example Neurons 3.3 Dynamics of Final-Goal Representation at the Population Level 4 Discussion References Dynamic Characteristics of Micro-state Transition Defined by Instantaneous Frequency in the Electroencephalography of Schizophrenia Patients 1 Introduction 2 Materials and Methods 2.1 Participants 2.2 EEG Recordings 2.3 Estimation of the Dynamic State Based on the Instantaneous Frequency Distribution 2.4 Statistical Analysis 3 Results 4 Discussion and Conclusions References Lagrange Programming Neural Networks for Sparse Portfolio Design 1 Introduction 2 Background 3 1-LPNN for Sparse Portfolio Selection 3.1 Development of 1-LPNN 3.2 Properties of the Sparse MV Problem and 1-LPNN 3.3 Global Stability 4 Experiments 4.1 Settling 4.2 Sparsity 4.3 Performance: Verification and Comparison 5 Conclusion References An Adaptive Convolution Auto-encoder Based on Spiking Neurons 1 Introduction 2 Methods 2.1 Neuron Model 2.2 Convolutional Spike Coding 2.3 Spike Pixel Value Mapping 2.4 Deep Convolutional Decoding 3 Experiment 3.1 DataSets 3.2 Determination of the Time Window 3.3 Determination of Coding Pre-training Parameters 3.4 Spike Convolutional Neural Networks (SCNNs) 4 Conclusion References Schizophrenia Detection Based on EEG Using Recurrent Auto-encoder Framework 1 Introduction 2 Materials and Methods 2.1 Dataset 2.2 Pre-processing 2.3 Methods 2.4 Baseline Methods 3 Experiments and Results 3.1 Model Implementation 3.2 Results 4 Discussion and Conclusion References Functional Roles of Amygdala and Orbitofrontal Cortex in Adaptive Behavior 1 Introduction 2 Model 2.1 Odor Discrimination Task and Reversal Task 2.2 Overview of the Model 2.3 The Model of ABL 2.4 The Model of OFC 2.5 Reinforcement Learning 2.6 Learning of ABL and OFC Networks 2.7 Two Trails After the Learning 3 Result 3.1 Learning of Adaptive Behavior 3.2 Formation of Odor-Taste Association in ABL and Expectation in OFC 3.3 Responses of OFC Networks for Two Task Trials 3.4 Reversal Learning of ABL and OFC 4 Conclusion References A Multiclass EEG Signal Classification Model Using Channel Interaction Maximization and Multivariate Empirical Mode Decomposition 1 Introduction 2 Material and Methodology 2.1 Dataset Details 2.2 Proposed Methodology 3 Results and Discussion 4 Conclusion and Future Scope References MVNet: Memory Assistance and Vocal Reinforcement Network for Speech Enhancement 1 Introduction 2 Related Work 2.1 Complex Structure of CRN 2.2 Speech Feature Information 3 Method 3.1 Memory Assistance 3.2 Vocal Reinforcement 3.3 Training Target 4 Experiment and Result 4.1 Experimental Setup 4.2 Ablation Study 4.3 Comprehensive Evaluation 5 Conclusions References Learning Associative Reasoning Towards Systematicity Using Modular Networks 1 Introduction 2 Related Work 3 Proposed Method 3.1 Tensor Product Representation 3.2 TPR-Based External Memory 3.3 Recurrent Encoder Modules 4 Experiments 4.1 Systematic Associative Recall Task 4.2 Concatenated-bAbI 5 Conclusion References Identifying Dominant Emotion in Positive and Negative Groups of Navarasa Using Functional Brain Connectivity Patterns 1 Introduction 2 Data Description 2.1 Stimuli Selection and Participants 2.2 EEG Data Recording and Preprocessing 3 Methodology 3.1 Functional Connectivity 3.2 Thresholding of Functional Networks 3.3 Network Metrics as Features 3.4 Machine Learning and Evaluation 4 Results 4.1 Identifying the Dominant Emotion in Positive and Negative Sets of Rasas Using Classification 4.2 Role of Functional Networks and Frequency Bands for Identifying the Significantly Distinguishable Pairs 4.3 Interpreting Outcome of Classifier Using Network Properties 5 Discussion 6 Conclusion References A Cerebellum-Inspired Model-Free Kinematic Control Method with RCM Constraint 1 Introduction 2 Preliminaries 2.1 Liquid State Machine 2.2 Manipulator Kinematics Model 2.3 Remote Center of Motion Constraint 3 Control Scheme Design 3.1 Cerebellum Model 3.2 Tracking Control Scheme 4 Simulations and Experiments 4.1 Simulations 4.2 Experiment 5 Conclusion References Instrumental Conditioning with Neuromodulated Plasticity on SpiNNaker 1 Introduction 2 Background 2.1 STDP and Synaptic Trace 2.2 Dopamine and Neuromodulated Plasticity 2.3 STDP and DA-modulation on SpiNNaker 3 Methodology and Results 3.1 Simulation Methods 3.2 The Balanced Random Network Parameterised for IZK Neurons 3.3 Neuromodulated Plasticity in the BRN 3.4 Instrumental Conditioning in the BRN 4 Conclusions References A Phenomenological Deep Oscillatory Neural Network Model to Capture the Whole Brain Dynamics in Terms of BOLD Signal 1 Introduction 2 Mathematical Model 2.1 Database Used 2.2 The Basic Model 2.3 1st Stage of Learning 2.4 2nd Stage of Learning 3 Results 4 Discussion 5 Conclusion and Future Goal References Explainable Causal Analysis of Mental Health on Social Media Data 1 Introduction 2 Background 3 Framework 3.1 Phase 1: Causal Categorization 3.2 Phase 2: Finding Explanations 3.3 Phase 3: Evaluations with Semantic Similarity 4 Experiments and Evaluation 4.1 CAMS Dataset 4.2 Experimental Setup 4.3 Experimental Results 4.4 Performance Evaluation 4.5 Ethical Considerations 5 Conclusion and Future Scope References Brain-Inspired Attention Model for Object Counting 1 Introduction 2 Method 3 Results 4 Conclusion and Future Work References Human Centered Computing Emotion Detection in Unfix-Length-Context Conversation 1 Introduction 2 Related Work 3 Our Method 3.1 Problem Formulation 3.2 Model 3.3 Training 4 Experiment Design 4.1 Dataset 4.2 Baselines 5 Experimental Result 5.1 Main Results 6 Conclusion References Multi-relation Word Pair Tag Space for Joint Entity and Relation Extraction 1 Introduction 2 Related Work 3 Proposed Approach 3.1 Task Definition 3.2 Tag Strategy and Decoding 3.3 Modeling Method 4 Experiments 4.1 Experimental Settings 4.2 Results and Analysis 4.3 Detailed Results on Overlapping Triples 4.4 The Model Efficiency 5 Conclusion References Efficient Policy Generation in Multi-agent Systems via Hypergraph Neural Network 1 Introduction 2 Preliminaries 2.1 Markov Game 2.2 Hypergraph Learning 3 Method 3.1 Hypergraph Generation 3.2 Hypergraph Convolution Critics 3.3 Learning with Hypergraph Convolution Critics 4 Experiments 4.1 Settings 4.2 Results and Analysis 5 Conclusion and Future Work References Bring Ancient Murals Back to Life 1 Introduction 2 Proposed Approach 2.1 Principle of Inpainting Network 2.2 Singular Value Decomposition (SVD) 2.3 Dense Spatial Attention with Mask 3 Experiments 3.1 Artificial Destruction Murals Repair Comparison 3.2 Experiment in Inpainting Real Damaged Murals 3.3 Ablation Study 4 Conclusion References Multi-modal Rumor Detection via Knowledge-Aware Heterogeneous Graph Convolutional Networks 1 Introduction 2 Related Work 2.1 Rumor Detection on Texts 2.2 Multi-modal Rumor Detection 3 Approach 3.1 Overview 3.2 Construction of Propagation Graph(CPG) 3.3 Cross-Modal Convergence of Knowledge (CCK) 3.4 Knowledge-Driven Graph Convolutional Network (KGCN) 3.5 Knowledgeable Debunker Classification (KDC) 4 Experimentation 4.1 Experimental Settings 4.2 Experimental Results 4.3 Ablation 4.4 Case Study 5 Conclusion References DAGKT: Difficulty and Attempts Boosted Graph-Based Knowledge Tracing 1 Introduction 2 Related Work 2.1 Knowledge Tracing 2.2 GIKT 3 The Proposed Model DAGKT 3.1 Framework 3.2 Embedding Module 3.3 Fusion Module 3.4 Knowledge Evolution Module and Prediction Module 4 Experiments 4.1 Setup 4.2 Overall Performance 4.3 Ablation Studies 5 Conclusion References AMRE: An Attention-Based CRNN for Manchu Word Recognition on a Woodblock-Printed Dataset 1 Introduction 2 Related Work 3 WMW Dataset Construction 3.1 Manchu Word Separation 3.2 Manchu Word Annotation 3.3 Dataset Statistic Analysis 4 AMRE 4.1 Feature Extraction Stage 4.2 Sequence Modeling Stage 4.3 Prediction Stage 5 Experiments 5.1 Evaluation Protocols 5.2 Results 5.3 Discussion 6 Conclusion References High-Accuracy and Energy-Efficient Action Recognition with Deep Spiking Neural Network 1 Introduction 2 Related Work 2.1 ANN-to-SNN Conversion 2.2 Recurrent Spiking Neural Network 2.3 Action Recognition 3 Methods 3.1 SNN Framework 3.2 RSNN Inference 3.3 Tandem Learning 4 Experiments 4.1 Experimental Setup 4.2 Implementation Details 4.3 Ablation Studies 4.4 Comparison with State-of-the-Art Methods 4.5 Energy Consumption 5 Conclusion References Sequence Recommendation Based on Interactive Graph Attention Network 1 Introduction 2 Methods 2.1 Problem Definition 2.2 User and Item Diagram Constructing Blocks 2.3 Information Interaction Module 3 Experiments 3.1 The Dataset and Its Sources 3.2 Parameter Setting 3.3 Correlation Comparison Model 3.4 Experimental Results and Analysis 3.5 Research on Node Dimension and Network Layers 4 Conclusion References Imbalanced Equilibrium: Emergence of Social Asymmetric Coordinated Behavior in Multi-agent Games 1 Introduction 2 Background 2.1 Multi-agent Partially Observable Markov Games 2.2 Policy Gradient, Actor-Critic and Proximal Policy Optimization 3 Proposed Method 3.1 Policy with Actions of Other Agents 3.2 Predictor Network 3.3 Predictor-Actor-Critic (PAC) 4 Experimental Evaluation 4.1 Experimental Setting 4.2 Comparative Methods 4.3 Performance Comparison and Analysis 4.4 Analyzing Learned Coordinated Behavior 5 Conclusion References Model-Based Reinforcement Learning with Self-attention Mechanism for Autonomous Driving in Dense Traffic 1 Introduction 2 Related Work 2.1 Attention Mechanism 2.2 Traffic Trajectory Prediction Methods 2.3 Reinforcement Learning 3 Methodology 3.1 Problem Definition 3.2 Environment Model 3.3 Model-Based RL for Dense Traffic 4 Experiments 4.1 Environment Setup 4.2 Environment Model Evaluation 4.3 Model-Based and Model-Free RL 5 Discussion 6 Conclusion References CLCDR: Contrastive Learning for Cross-Domain Recommendation to Cold-Start Users 1 Introduction 2 The Proposed CLCDR Framework 2.1 Notations 2.2 Overview of CLCDR 3 Experiments 3.1 Experimental Settings 3.2 Experimental Results and Analysis 4 Conclusion References Medical Visual Question Answering via Targeted Choice Contrast and Multimodal Entity Matching 1 Introduction 2 Method 2.1 The Proposed Med-VQA Framework 2.2 Targeted Choice Contrast (TCC) 2.3 Multimodal Entity Matching (MEM) 3 Experiments 3.1 Dataset 3.2 Experimental Setup 3.3 Comparison with the State-of-the-arts 3.4 Ablation Studies 3.5 Comparison of Contrastive Learning Methods 3.6 Superiority of MEM Module 3.7 Visualization 4 Conclusion References Boosting StarGANs for Voice Conversion with Contrastive Discriminator 1 Introduction 2 Background 2.1 StarGAN-VC2 Method 2.2 Simple Siamese Representation Learning 3 Methodology 3.1 Contrastive Learning for Real Samples 3.2 Supervised Contrastive Learning for Fake Speech Samples 4 Experiments 4.1 Experimental Setup 4.2 Objective Evaluation 4.3 Subjective Evaluation 4.4 Training Stability 4.5 Ablation Study on Contrastive Losses 5 Conclusion References Next POI Recommendation with Neighbor and Location Popularity 1 Introduction 2 Related Work 2.1 General POI Recommendation 2.2 Successive POI Recommendation 3 Preliminaries 3.1 Problem Formulation 4 Approach 4.1 Multimodal Embedding Module 4.2 Neighbor Discovery Module 4.3 Self-attention Module 4.4 Location Popularity Module 4.5 Prediction Layer Module 5 Experiments 5.1 Datasets and Parameter Setting 5.2 Evaluation Metrics 5.3 Baseline Approach 5.4 Results and Analysis 6 Conclusion References Trustworthiness and Confidence of Gait Phase Predictions in Changing Environments Using Interpretable Classifier Models 1 Introduction 1.1 Related Work 2 Generalized Matrix Learning Vector Quantization as an Interpretable Machine Learning Classifier 3 Experimental Setup 3.1 The Collected Data 3.2 Feature Generation 3.3 Model Training 3.4 Postprocessing and Evaluation 4 Results 5 Conclusion References Enhance Gesture Recognition via Visual-Audio Modal Embedding 1 Introduction 2 Proposed Approach 2.1 Model Architecture 2.2 Multimodal Collaborative Representation 2.3 Multimodal Joint Training 3 Experiments 3.1 Dataset 3.2 Preliminary Experiments 3.3 Multimodal Embedding Experiments 3.4 Visualization of Gesture Embedding 4 Conclusion References Learning from Fourier: Leveraging Frequency Transformation for Emotion Recognition 1 Introduction 2 The Proposed Method 2.1 Pipeline of FATENet 2.2 Optional Structure and Loss Function 3 Experimental Results 3.1 Dataset 3.2 Compared Methods 3.3 Main Results 3.4 Ablation Studies 3.5 Analysis 4 Conclusion References Efficient Double Oracle for Extensive-Form Two-Player Zero-Sum Games 1 Introduction 2 Background 3 Related Work 4 Method 4.1 Efficient Double Oracle 4.2 Neural Efficient Double Oracle 5 Experiments 5.1 Experiments with Tabular Methods 5.2 Experiments with Deep Reinforcement Learning 6 Conclusion References FastThaiCaps: A Transformer Based Capsule Network for Hate Speech Detection in Thai Language 1 Introduction 2 Related Works 2.1 Works on English Data 2.2 Works on Thai Data 3 Dataset Description 4 Methodology 4.1 Text Embedding Generation 4.2 Proposed FastThaiCaps Model 4.3 Channel-1 4.4 Channel-2 4.5 Loss Function 5 Experimental Results and Analysis 5.1 Baseline Setup 5.2 Findings from Experiments 5.3 Error Analysis 6 Conclusion and Future Work References Author Index The four-volume set CCIS 1791, 1792, 1793 and 1794 constitutes the refereed proceedings of the 29th International Conference on Neural Information Processing, ICONIP 2022, held as a virtual event, November 22-26, 2022.