Knowledge Science, Engineering and Management: 16th International Conference, KSEM 2023, Guangzhou, China, August 16–18, 2023, Proceedings, Part I (Lecture Notes in Artificial Intelligence)

This volume set constitutes the refereed proceedings of the 16th International Conference on Knowledge Science, Engineering and Management, KSEM 2023, which was held in Guangzhou, China, during August 16–18, 2023. The 114 full papers and 30 short papers included in this book were carefully reviewed and selected from 395 submissions. They were organized in topical sections as follows: knowledge science with learning and AI; knowledge engineering research and applications; knowledge management systems; and emerging technologies for knowledge science, engineering and management. Preface Organization Keynotes Abstracts Credibility of Machine Learning Through Information Granularity A New Paradigm to Leverage Formalized Knowledge and Machine Learning Recent Advances in Assessing Time Series Similarity Through Dynamic Time Warping ChatGLM: Run Your Own “ChatGPT” on a Laptop Contents – Part I Knowledge Science with Learning and AI Joint Feature Selection and Classifier Parameter Optimization: A Bio-Inspired Approach 1 Introduction 2 Related Work 3 Algorithm Description 3.1 Honey Badger Algorithm 3.2 IBHBA 3.3 Complexity Analysis of IBHBA 3.4 Feature Selection Based on IBHBA 4 Experiment and Analysis 4.1 Experiment Setup 4.2 Results and Analysis 5 Conclusions References Automatic Gaussian Bandwidth Selection for Kernel Principal Component Analysis 1 Introduction 2 Methodology 2.1 Preliminaries 2.2 Criterion of the Maximum Sum of Eigenvalues (CMSE) Method 2.3 Scalable CMSE (SCMSE) Method 2.4 Choosing the Number of Principal Components 3 Experiment 4 Efficiency of SCMSE 4.1 Ten-Dimensional Hyperspheres 4.2 A Real Large-Scale Data Example: Letter Recognition Data Set 5 Discussion References Boosting LightWeight Depth Estimation via Knowledge Distillation 1 Introduction 2 Related Work 3 Lightweight Network 4 Promoting KD with Auxiliary Data 5 Experiments 5.1 Implementation Details 5.2 Quantitative Evaluation 5.3 Comparison with Previous Methods 5.4 Effect of Varying the Number of Auxiliary Data 5.5 Cross-Dataset Evaluation 6 Conclusion References Graph Neural Network with Neighborhood Reconnection 1 Introduction 2 Notation and Preliminary 3 Network Reconnection 4 The Framework of NRGNN 5 Experiments 5.1 Experimental Setting 5.2 Performance on Classical Datasets 5.3 Performance on Twitch Datasets 5.4 Ablation Study 6 Conclusion References Critical Node Privacy Protection Based on Random Pruning of Critical Trees 1 Introduction 2 CNPP Model 3 Scheme 3.1 Node Criticality Calculation 3.2 Critical Structure Extraction 3.3 Random Pruning Based on Critical Tree Privacy Protection (RPCT) 4 Experiment 4.1 Experiment Preparation 4.2 Global Network Analysis 4.3 Clustering Analysis 5 Conclusions References DSEAformer: Forecasting by De-stationary Autocorrelation with Edgebound 1 Introduction 2 Model 2.1 Encoder 2.2 Decoder 2.3 Sequence Preprocessing 2.4 De-stationary Autocorrelation(DSA) 2.5 Edgebound 3 Experiments 3.1 Datasets and Evaluation Metrics 3.2 Baselines 3.3 Results 4 Conclusion References Multitask-Based Cluster Transmission for Few-Shot Text Classification 1 Introduction 2 Environment Setting 2.1 Task Definition 2.2 Encoder 3 Algorithm and Architecture 3.1 Text Encoding 3.2 Cluster Transmission 3.3 Loss Calculation 3.4 Auxiliary Learning 4 Experiment 4.1 Experimental Conditions and Dataset 4.2 Results Analysis 4.3 Visualization Analysis 4.4 Fitting Verification 5 Conclusion References Hyperplane Knowledge Graph Embedding with Path Neighborhoods and Mapping Properties 1 Introduction 2 Related Work 2.1 Translation-Based Embedding Methods 2.2 Muti-source Information Learning 3 Methodology 3.1 TransPMH 3.2 Objective Formalization 4 Experiments 4.1 Dataset 4.2 Link Prediction 4.3 Triplet Classification 4.4 Auxiliary Experiment 5 Conclusion References RTAD-TP: Real-Time Anomaly Detection Algorithm for Univariate Time Series Data Based on Two-Parameter Estimation 1 Introduction 2 Related Work 3 Methodology 3.1 Initial Sequence Modeling Module 3.2 GPD Parameter Estimation Module 3.3 Threshold Calculation Module 3.4 Anomaly Detection Module 3.5 Data Smoothing Module 4 Experiments 4.1 Datasets 4.2 Metrics 4.3 Baselines 4.4 Experiment 1 4.5 Experiment 2 5 Conclusion References Multi-Sampling Item Response Ranking Neural Cognitive Diagnosis with Bilinear Feature Interaction 1 Introduction 2 Preliminaries 2.1 Item Response Ranking (IRR) 2.2 Bilinear Feature Interaction 3 Proposed Model 3.1 The Design of Farmework 3.2 Multi-Sampling Item Response Ranking 4 Experiments 4.1 Experiment Platform and Datasets 4.2 Evaluation Indicators and Baselines 4.3 Experimental Results and Analysis 5 Conclusion References A Sparse Matrix Optimization Method for Graph Neural Networks Training 1 Introduction 2 Sparse Matrix Format and Performance Model 2.1 Overview 2.2 BMCOO Storage Format for Graphs 2.3 SpMV Performance Model for Graph Partitioning 3 Experiments 3.1 Datasets 3.2 Experiment Setting 3.3 Results and Analysis 4 Conclusion References Dual-Dimensional Refinement of Knowledge Graph Embedding Representation 1 Introduction 2 Related Work 3 Method 3.1 Entity Level 3.2 Relation Level 3.3 Model Computing and Training 4 Experiments 4.1 Experimental Setup 4.2 Link Prediction 4.3 Relation Prediction 4.4 Ablation Studies 5 Conclusion References Contextual Information Augmented Few-Shot Relation Extraction 1 Introduction 2 Related Work 3 Proposed Method 3.1 Sentence Augmentation Method by Relation Information 3.2 Relation Prototype Representation 3.3 Loss Function 3.4 Relation Extraction 4 Experiments 4.1 Dataset 4.2 Training and Validation 4.3 Comparable Models 4.4 Results 4.5 Result Analysis 4.6 Ablation Study 4.7 Limitation 5 Conclusion References Dynamic and Static Feature-Aware Microservices Decomposition via Graph Neural Networks 1 Introduction 2 Related Work 3 Proposed Approach 3.1 Approach Overview 3.2 Problem Formulation 3.3 Dynamic Analysis 3.4 Static Analysis 3.5 Construction of Dynamic and Static Features Graph 3.6 Training and Clustering 4 Evaluation 4.1 Evaluation Applications 4.2 Baselines 4.3 Metrics 4.4 Results and Discussion 5 Conclusions References An Enhanced Fitness-Distance Balance Slime Mould Algorithm and Its Application in Feature Selection 1 Introduction 2 Slime Mould Algorithm 3 The Proposed EFDB-SMA Algorithm 3.1 Elite Opposition-Based Learning 3.2 Roulette-Wheel Fitness-Distance Balance 3.3 Chaotic Perturbation 3.4 Greedy Selection Strategy 4 Experimental Results and Analysis 4.1 Parameter Settings 4.2 Comparison with Other Algorithms 5 Application in Feature Selection 5.1 Feature Selection 5.2 Binary EFDB-SMA 5.3 Datasets and Parameters 5.4 Experimental Results and Analysis 6 Conclusion References Low Redundancy Learning for Unsupervised Multi-view Feature Selection 1 Introduction 2 Multi-view Feature Selection with Low Redundancy Learning 2.1 Notations 2.2 Formulation 2.3 Optimization 2.4 Convergence Analysis 3 Experiments 4 Conclusion References Dynamic Feed-Forward LSTM 1 Introduction 2 Related Work 3 Model 4 Experiments 4.1 Datasets 4.2 Competitor Models 4.3 Ablation Study 4.4 The Statistic of Classification Task 4.5 The Statistic of Sequence Labeling Task 5 Conclusion References Black-Box Adversarial Attack on Graph Neural Networks Based on Node Domain Knowledge 1 Introduction 2 Related Work 3 Preliminary 3.1 Notations 3.2 Graph Neural Network 3.3 The Graph Adversarial Attack Settings 4 Methodology 4.1 Attack Model 4.2 Node Selection 4.3 Feature Selection 4.4 The Setting of the Feature Perturbation Vector 4.5 Topology Attack 5 Experiments 5.1 Experimental Setting 5.2 Effectiveness Evaluation 5.3 Ablation Studies 6 Conclusion and Future Works References Role and Relationship-Aware Representation Learning for Complex Coupled Dynamic Heterogeneous Networks 1 Introduction 2 Preliminaries 3 The Proposed Model: RRDNE 3.1 Construct a Historical Memory Graph 3.2 Random Walk Strategy Based on Role Perception 3.3 Improved Skip-Gram Model Bases on Relational Awareness 3.4 The RRDNE Model 4 Experiments 4.1 Evaluation Datasets 4.2 Experiment Settings 4.3 Node Classification 4.4 Link Prediction 4.5 Parametric Analysis 5 Conclusions References Twin Graph Attention Network with Evolution Pattern Learner for Few-Shot Temporal Knowledge Graph Completion 1 Introduction 2 Related Work 3 Background Knowledge 4 Model 4.1 Twin Graph Attention Network for entities 4.2 DA: Query Time-Difference Pair Encoder 4.3 VE: Evolution Pattern Encoder for Few-Shot Instances 4.4 Semantic Decode for Prediction 5 Experiments 5.1 Datasets 5.2 Baselines 5.3 Implementation Details 5.4 Experimental Results 5.5 Ablation Study 5.6 Impacts of in DA 6 Conclusion References Subspace Clustering with Feature Grouping for Categorical Data 1 Introduction 2 Related Work 3 Subspace Clustering with Feature Grouping Strategy 3.1 Initial Clustering 3.2 Construction of Feature-to-Cluster Groups 3.3 Local Clustering 3.4 Global Clustering 4 Experiments 5 Conclusion References Learning Graph Neural Networks on Feature-Missing Graphs 1 Introduction 2 Methodology 2.1 Notation 2.2 Feature Information Completion 2.3 Alignment Mechanisms 2.4 Relation Constraint Mechanism 2.5 Model Objective 3 Experiments 3.1 Experimental Setup 3.2 Node Classification 4 Conclusion References Dealing with Over-Reliance on Background Graph for Few-Shot Knowledge Graph Completion 1 Introduction 2 Related Work 3 Proposed Framework 3.1 Meta-relation Learner 3.2 Analogical Enhancer 3.3 Score Learner 3.4 Optimization 4 Experimental Results 4.1 Dataset and Baseline 4.2 Implementation Details and Evaluation Metrics 4.3 Main Results 4.4 Ablation Study 5 Conclusion References Kernel-Based Feature Extraction for Time Series Clustering 1 Introduction 2 Related Work 3 Isolation Kernel-Based Time Series Encoding 3.1 Detection Feature Points and Describing Representative Subsequences 3.2 Encoding Time Series 4 Empirical Evaluation 5 Conclusion References Cluster Robust Inference for Embedding-Based Knowledge Graph Completion 1 Motivation, Objective and Related Work 1.1 Motivation and Objective 1.2 Related Work 2 Theoretical Background 2.1 Graphs, KGs, KG Clustering and KGC 2.2 KGC as a Semi-supervised ML Task 3 Cluster-Robust Inference for KGC 3.1 Problem Statement 3.2 Suggested Approach 4 Experiments 4.1 Experimental Setup 4.2 Results 5 Conclusion, Discussion and Outlook 5.1 Conclusion 5.2 Discussion and Outlook References Community-Enhanced Contrastive Siamese Networks for Graph Representation Learning 1 Introduction 2 Related Work 3 Method 3.1 Graph Augmentations 3.2 Siamese Graph Network 3.3 Deep Clustering Optimization 3.4 Debiased Cross Contrast 3.5 MEDC Algorithm 4 Experiments and Analysis 4.1 Experimental Settings 4.2 Results and Analysis 4.3 Parameter Sensitivity Analysis 5 Conclusion References Distant Supervision Relation Extraction with Improved PCNN and Multi-level Attention 1 Introduction 2 Related Work 3 Relation Extraction with I-PCNN and Multi-level Attention 3.1 Sentence Encoder 3.2 Selective Attention over Sentences 3.3 Selective Attention over Bags 4 Experiments and Analysis 4.1 Data Set and Evaluation Metrics 4.2 Configurations 4.3 Comparison with Baseline Models 4.4 Effect of I-PCNN 5 Conclusion References Enhancing Adversarial Robustness via Anomaly-aware Adversarial Training 1 Introduction 2 Related Work 2.1 Adversarial Attack and Defense 2.2 Adversarial Training 3 Problem Formulation 3.1 Notations 3.2 Formulation of Adversarial Training 3.3 Formulation of Normal/Abnormal Cases 3.4 Our Solution 4 Method 4.1 Anomaly-aware Learning Strategy 4.2 Anomaly-aware Case Indicator 4.3 Anomaly-aware Adversarial Training Framework 5 Experimental Results 5.1 Experimental Setup 5.2 Performance Analysis 5.3 Performance on Tiny ImageNet 5.4 Ablation Studies and Additional Discussion 6 Conclusion References An Improved Cross-Validated Adversarial Validation Method 1 Introduction 2 Our Proposed Improved Adversarial Validation 2.1 Formalization of an Adversarial Validation 2.2 Our Proposed Improved Adversarial Validation 3 Investigation of the Existing Play-and-Plug Algorithm Comparison Methods 4 Experiments 4.1 Experimental Data Sets and Settings 4.2 Experimental Results and Analysis 5 Conclusions References EACCNet: Enhanced Auto-Cross Correlation Network for Few-Shot Classification 1 Introduction 2 Related Work 2.1 Few-Shot Classification 2.2 Auto-Correlation 2.3 Cross-Correlation 3 Proposed Approach 3.1 Preliminary on Few-Shot Classification 3.2 Architecture Overview of EACCNet 3.3 Enhanced Auto-Correlation Representation (EACR) 3.4 Enhanced Cross-Correlation Attention (ECCA) 3.5 Training Strategy of EACCNet 4 Experiments 4.1 Datasets 4.2 Implementation Details 4.3 Experimental Results 4.4 Ablation Studies 5 Conclusion References A Flexible Generative Model for Joint Label-Structure Estimation from Multifaceted Graph Data 1 Introduction 2 Background 2.1 Graph Neural Networks for Semi-supervised Node Classification 2.2 Graph Structure Learning 3 Problem Statement 4 Our Proposed Model 4.1 Probabilistic Bayesian Model 4.2 Model Instantiations 4.3 Model Training 4.4 Computational Complexity 5 Experiments 5.1 Experimental Settings 5.2 Experimental Results 5.3 Conclusion References Dual Channel Knowledge Graph Embedding with Ontology Guided Data Augmentation 1 Introduction 2 Related Work 3 Ontology-Guided Data Augmentation 4 Joint Embedding Framework 4.1 Dual Channel Data Augmentation 4.2 Joint Loss 4.3 Empirical Evaluation Metrics on C-axiom and V-axiom 5 Experiments 5.1 Experimental Setup 5.2 Result and Discussion 5.3 Case Study 6 Conclusion References Multi-Dimensional Graph Rule Learner 1 Introduction 2 Related Work 3 Definitions 4 Proposed Algorithm 5 Experimental Evaluation 5.1 Datasets 5.2 Path Transfer Probability (F II) 5.3 Embedding Distance (F III) 5.4 The Prediction Performance of MDGRL (Multi-F) 6 Conclusion References MixUNet: A Hybrid Retinal Vessels Segmentation Model Combining The Latest CNN and MLPs 1 Introduction 2 Method 2.1 Encoder-Decoder 2.2 Mutil-scale Vision Attention Module 2.3 Mix-Scale MLP 2.4 Global Context Path 3 Experiment 3.1 Dataset Introduction and Amplification 3.2 Experimental Environment 3.3 Experimental Results and Analysis 4 Conclusion References Robust Few-Shot Graph Anomaly Detection via Graph Coarsening 1 Introduction 2 Related Work 2.1 Graph Anomaly Detection 2.2 Meta-Learning on Graphs 3 Framework 3.1 Problem Definition 3.2 Graph Embedding with Graph Coarsening 3.3 Loss Function 3.4 Meta-Learning for Networks 3.5 Anomaly Detection Using RCM-GAD 4 Experiments and Results 4.1 Experimental Setup 4.2 Graph-Level Anomaly Detection 4.3 Subgraph-Level Anomaly Detection 4.4 Ablation Experiments 4.5 Effectiveness of Graph Coarsening Module 5 Conclusion References An Evaluation Metric for Prediction Stability with Imprecise Data 1 Introduction 2 Related Work 3 Stability 3.1 Stability of a Sample 3.2 Stability of a Model 4 Experiments 4.1 Datasets 4.2 Implementation Details 4.3 Result and Analysis 5 Conclusion References Reducing the Teacher-Student Gap via Elastic Student 1 Introduction 2 Related Work 2.1 Knowledge Distillation 2.2 Re-parameterization 3 Method 3.1 Elastic Architecture 3.2 Elastic Learning 4 Experiments 4.1 Main Results 4.2 Ablation Study 5 Conclusion References Author Index