Artificial Intelligence in China: Proceedings of the 4th International Conference on Artificial Intelligence in China

Description

This book brings together papers presented at the 4th International Conference on Artificial Intelligence in China (ChinaAI), Changbaishan, China, on July 23-24, 2022, which provides a venue to disseminate the latest developments and to discuss the interactions and links between these multidisciplinary fields. Spanning topics covering all topics in Artificial Intelligence with new development in China, this book is aimed at undergraduate and graduate students in Electrical Engineering, Computer Science and Mathematics, researchers and engineers from academia and industry as well as government employees (such as NSF, DOD, DOE, etc). Contents Traffic Flow Prediction Based on ACO-BI-LSTM 1 Introduction 2 Ant Colony Optimization Algorithms 2.1 State Transition Rules 2.2 Pheromone Update Rules 3 Bi-LSTM Model 4 ACO-Bi-LSTM Forecasting Model 4.1 Model Framework 4.2 Ant Colony Algorithm Optimizes Model Parameters 4.3 Model Training 5 Experimental Analysis 6 Data Preprocessing 7 Model Evaluation 8 Analysis of Results 9 Conclusion References Research on Natural Language Processing Technology in Construction Project Management 1 Introduction 2 Introduction to Natural Language Processing 3 Application of NLP Technology 3.1 Application of Risk in Construction Contract 3.2 Visual Analysis of Building Safety Knowledge Atlas 3.3 Research on BIM Visualization Based on Knowledge Map 4 Conclusion References Exploring the Application of SketchUp and Twinmotion in Building Design Planning 1 Introduction 2 Modeling Tools - SketchUp 3 Real-Time Rendering Tool - Twinmotion 3.1 Introduction to Twinmotion 3.2 Research on the Specific Function of Twinmotion 3.3 Real-Time Interconnection Between SketchUp and Twinmotion Data 3.4 Unreal Engine Detailed Design 4 The Application of SketchUp and Twinmotion in the Project 4.1 Local Regional Planning of Huadian City, Jilin Province 4.2 Architectural Design and Planning of a Commercial Center in Changchun 4.3 Architectural Design and Planning of a Commercial Center in Changchun 5 Conclusion References Lifecycle-Based Software Defect Prediction Technology 1 Introduction 2 Analysis of Software Prediction Technology 2.1 Definition of Software Prediction Technology 2.2 Principles of Software Prediction Technology 2.3 Main Workflow of Software Prediction Technology 3 Software Reliability Measurement 4 Lifecycle-Based Software Defect Prediction Technology 4.1 Existing Research 4.2 Experimental Analysis 4.3 Index Analysis 4.4 Method Analysis 5 Software Defect Prediction Technology Defect Report Analysis 6 Conclusion References Unsupervised Anomaly Detection Method Based on DNS Log Data 1 Introduction 2 Anomaly Detection Based on DNS Log Data 2.1 Feature Extraction 2.2 Anomaly DNS Behavior Detection 3 Experiment and Result Analysis 3.1 Performance Validation of the Detection Model 3.2 Performance Verification of Blacklist and Whitelist 4 Conclusion References Binary Hyperdimensional Computing for Image Encoding 1 Introduction 2 Hyperdimensional Computing 2.1 Addition (Bundling) 2.2 Multiplication (Binding) 2.3 Permutation 3 Proposed Method 4 Experiment 4.1 Setup 4.2 Experimental Results 5 Conclusion References Application of Artificial Intelligence in Water Supply Dispatching of Smart City 1 Introduction 2 Conventional Intelligent Control and Early Warning Function of Urban Water Supply 2.1 Overall Planning and Intelligent Dispatching of Urban Water Supply Facilities 2.2 Accurate and Intelligent Dispatching of Urban Water Supply Network 2.3 Intelligent Control of Urban Water Supply Leakage 3 Intelligent Application in Urban Water Supply Emergency 3.1 Emergency Intelligent Control of Urban Water Supply Facilities 3.2 Emergency Intelligent Control of Urban Water Supply Network 4 Conclusion References Remote Sensing Image Object Detection Based on Improved SSD Algorithm 1 Introduction 2 Improved SSD Algorithm 2.1 Overall Network Structure Design 2.2 SimAm Attention Mechanism 2.3 Decoupling Head 3 Materials and Methods 3.1 RSOD Dataset 3.2 Evaluating Indicator 4 Experiment and Result Analysis 4.1 Experimental Environment 4.2 Comparative Test of Different Algorithm 5 Conclusion References Detection Method of Aggregated Floating Objects on Water Surface Based on Attention Mechanism and YOLOv3 1 Introduction 2 YOLOv3 Model Improvement 2.1 Anchors Presetting Based on K-means++ algorithm 2.2 Localization Loss Function Improvement 3 S-YOLOv3 Network 3.1 SE Attention Module 3.2 S-YOLOv3 Network Structure 4 Experimental Results and Analysis 4.1 Dataset Preparation 4.2 Experimental Environment and Conditions 4.3 Experiments and Analysis of Results 5 Conclusions References Research on the Text2SQL Method Based on Schema Linking Enhanced 1 Introduction 2 Related Work 3 Algorithm Description 3.1 Schema Linking Enhanced 3.2 Encoder-Decoder 4 Experiments and Analysis 4.1 Dataset and Evaluating Indicator 4.2 Experiment Setting 4.3 Experimental Results and Analysis 5 Summary References A Multi-task Learning Model for Emotion-Cause Extraction Based on Emotion Classification 1 Introduction 2 Related Work 3 Model 3.1 Problem Description 3.2 Model Framework 3.3 Position Feature Coding 3.4 Focal Loss Function 3.5 Training Objectives 4 Experimental Analysis 4.1 Dataset Analysis 4.2 Experimental Parameter Setting 4.3 Experimental Results and Analysis 5 Summarize References Research on Helmet Detection Algorithm Based on Improved YOLOv5s 1 Introduction 2 Related Algorithms 2.1 YOLOv5 Algorithm Principle 2.2 YOLOv5s Algorithm Principle 3 Algorithm Improvement 3.1 GhostNet 3.2 ECA-Net 4 Experimental Results and Analysis 4.1 Experimental Environment Configuration and Data Introduction 4.2 Evaluation Indicators 4.3 Analysis of Experimental Results 5 Conclusion References Label Embedding Based Scoring Method for Secondary School Essays 1 Introduction 2 Related Work 3 Method 3.1 Feature Extraction Module 3.2 Feature Fusion Module 3.3 Essay Scoring Module 4 Experiments and Analysis of Results 4.1 Data Sets and Evaluation Metrics 4.2 Experimental Configuration 4.3 Analysis of Results 4.4 Feature Extraction Layer Analysis 4.5 Feature Fusion Layer Analysis 5 Conclusion References Research on Wheat Ears Detection Method Based on Improved YOLOv5 1 Introduction 2 Problems with YOLOv5 in the Detection of Wheat Ears 3 Wheat-YOLOv5 Detection Algorithm 3.1 Fusion ECA Effective Channel Attention Mechanism 3.2 Introduction of Separation-Based Convolution 3.3 Improved Loss Function 4 Experimental Results and Analysis 4.1 Experimental Dataset 4.2 Experimental Environment and Parameters 4.3 Experimental Analysis 5 Summary References Aircraft Target Detection Algorithm Based on Improved YOLOv5s 1 Introduction 2 Related Algorithms 2.1 YOLOv5 Algorithm 2.2 RepVGG 2.3 GAMAttention 3 Algorithm Improvement 4 Experimental Results and Analysis 4.1 Experimental Environment Configuration and Data Introduction 4.2 Ablation Experiment and Result Analysis 4.3 RG-YOLOv5 Compared with Other Algorithms 5 Conclusion References A Survey in Virtual Image Generation Based on Generative Adversarial Networks 1 Introduction 2 GAN Preliminaries 3 General Approaches of Virtual Scene Generation 3.1 Direct Method 3.2 Staged Method 4 Summary References Research on the Service for the Disabled in Interior Design and Product Design Based on Artificial Intelligence 1 Introduction 2 Application Analysis of Artificial Intelligence in Interior Design and Product Design 3 Analysis of the Disabled and Special Groups 4 The Application of Artificial Intelligence in Residence 4.1 Application Research of Artificial Intelligence in Interior Space Design 5 Application of Artificial Intelligence in Product Design for the Disabled 5.1 Proposed Framework for Face Detection and Recognition in CCTV Images 5.2 Intelligent Monitoring 5.3 Intelligent Voice Prompt 5.4 Intelligent Security System 6 Conclusion References Application and Prospect of Artificial Intelligence Image Analysis Technology in Natural Resources Survey 1 Introduction 2 Deep Learning 2.1 Convolutional Neural Network 2.2 Fully Convolutional Neural Networks 3 Application and Prospect of AI Image Analysis Technology in Natural Resources Survey and Monitoring 3.1 LSTM Networks Model 3.2 Application Prospect 4 Summary References Application and Technical Analysis of Computer Vision Technology in Natural Resource Survey 1 Introduction 2 Computer Vision Technology 2.1 Image Classification 2.2 Object Detection 2.3 Semantic Segmentation 3 Application of CV Technology in Natural Resource Survey 3.1 Remote Sensing Image 3.2 Wildlife Monitoring 4 Conclusion References Review of Generative Adversarial Networks in Object Detection 1 Introduction 2 GAN and Its Derivative Models in the Field of Object Detection 2.1 Perpetual GAN 2.2 MTGAN 2.3 CGAN 3 GAN’S Application in the Field of Object Detection 3.1 Industrial Defect Detection 3.2 Medical Image Detection 3.3 Remote Sensing Image Detection 3.4 Face Detection 4 Conclusion References A Review of the Application of Convolutional Neural Networks in Object Detection 1 Introduction 2 Application of Object Detection Based on Candidate Region 2.1 R-CNN 2.2 SPPNet 2.3 Fast R-CNN 2.4 Faster R-CNN 2.5 R-FCN 2.6 Mask R-CNN 2.7 Cascade R-CNN 2.8 TridentNet 3 Application of Object Detection Based on Regression 3.1 YOLO Series 3.2 SSD Series 4 Video Object Detection Application 5 Conclusion References Optimal Deployment of Large-Scale Wireless Sensor Network Based on Topological Potential Adaptive Graph Clustering 1 Introduction 2 Theory and Algorithm 2.1 Adaptive Graph Clustering Based on Topological Potential 2.2 Optimal Nodes Selection Based on SVD-QR 3 Experiment 3.1 Dataset and Parameter Settings 3.2 Experimental Results and Analysis 4 Conclusion References INSL: Text2SQL Generation Based on Inverse Normalized Schema Linking 1 Introduction 2 Related Work 3 Definition 4 Database Schema Inverse Normalization 4.1 Basic Model 4.2 Inverse Normalization 4.3 Schema Linking 5 Experiments and Analysis of Results 5.1 Datasets 5.2 Evaluation Indicators 5.3 Quality Assessment 6 Summary References The Research on Prediction for Financial Distress in Car Company Listed Combining Financial Indicators and Text Data 1 Introduction 2 Methodology 2.1 Financial Text Feature Extraction Model 2.2 Non-financial Text Sentiment Indicator Model 2.3 Financial Distress Forecasting Model 3 Experiment and Analysis 3.1 Data Preparation 3.2 Experiment I - Based on Financial Indicator Data 3.3 Experiment II - Based on Financial Indicator Data, Financial Text Data and Non-financial Text Data 4 Conclusion and Future Works References Frame Interpolation for Weather Radar Data 1 Introduction 2 Related Work 2.1 Data Processing Methods for Weather Radar Data 2.2 Video Frame Interpolation 2.3 Depth Estimation 3 The Proposed Approach 3.1 Algorithm Overview 3.2 Flow Estimation 3.3 Depth Estimation 3.4 Context Extraction 3.5 Implimentation Details 4 The Experiments 5 Conclusion References A Hybrid Neural Network for Music Generation Using Frequency Domain Data 1 Introduction 2 Related Works and Preliminaries 2.1 Music Data Representation 2.2 Discrete Short-Time Fourier Transform 3 Model and Formulation 3.1 Data Preprocessing 3.2 Model 3.3 Griffin-Lim Algorithm (GLA) 4 Experiment and Evaluation 4.1 Low-Level Feature Comparison 5 Conclusion References Prediction Model for Inclusive Finance Development Considering the Impact of COVID-19: The Case of China 1 Introduction 2 Methodology and Related Theories 2.1 Bag of Words Model TF-IDF Algorithm 2.2 Principles of Random Forest Regression 2.3 Methodology 3 Experiment and Analysis 3.1 Data Preparation 3.2 Experiment I – Data Augmentation Model 3.3 Experiment II – Text Prediction Model 4 Conclusion and Future Works References Certifiable Optimal Visual Pose Estimation for Space Applications 1 Introduction 2 Certifiable Optimal Relative Pose Estimation 3 Certifiable Optimal Pose-Graph Optimization 4 Conclusions References Automotive Manufacturing Revenue Prediction Using Financial and Comment Sentiment Data Based on CNN Model 1 Introduction 2 Methodology 3 Experiment and Analysis 3.1 Data Preparation 3.2 Experiment I - Based on Random Forest 3.3 Experiment II - with Enterprise’s Own Indicators Only 3.4 Experiment III - Without Comment Variables 3.5 Experiment IV – Proposed Model 4 Conclusion and Future Works References Deep Learning-Enhanced ICA Algorithm for Sub-Gaussian Blind Source Separation 1 Introduction 2 Enhanced ICA Network 2.1 From Math Iterative Operations to Augmented Networks 2.2 Structural Design and the Pre-training of Explainable Neural Networks 2.3 Enhanced ICA Network Embedded in Deep Learning 3 Simulations 3.1 Training Process and the Result 3.2 The Effect of the Enhanced ICA Network 4 Conclusion References Research on Semantic Verification Method of AIXM Data Based on SBVR 1 Introduction 2 Data Verification Method Summary 2.1 Overview of AIXM Data Verification Method 2.2 Semantics of Business Vocabulary and Rules (SBVR) 2.3 Business Rule Language Analysis 3 AIXM Semantic Rule Meta Model Based on SBVR 3.1 AIXM Business Rule Classification 3.2 AIXM Semantic Rule Meta Model 4 AIXM Rule Writing Method and Coding 4.1 AIXM Rule Writing Method 4.2 Application of Business Rule Coding 5 Conclusion References A Clustering-Based Anomaly Detection for Unstable Approach in Terminal Airspace 1 Introduction 2 Data Sets and Pre-processing 3 Air Traffic Flow and Spatial Anomalies 4 Energy Management 4.1 Energy Safety Boundary 4.2 Atypical Scores 5 Conclusions References Research on Pedestrian Detection and Recognition Based on Improved YOLOv6 Algorithm 1 Introduction 2 YOLO Algorithm and Attention Mechanism 2.1 YOLO 2.2 SE Attention Module 2.3 The Improved Network Structure 2.4 Experimental Results and Analysis References Jetson Nano-Based Pedestrian Density Detection in Subway Stations 1 Introduction 2 Theoretical Basis 2.1 Convolutional Neural Network 3 Yolov4-Tiny Algorithm-Based Pedestrian Detection Counting 3.1 Introduction of the Yolov4-Tiny Algorithm 3.2 Evaluation Criteria 3.3 Experimental Environment and Parameter Settings 3.4 Experimental Results 4 Embedded Jetson Nano-Based Human Flow Density Detection Implementation 4.1 The Choice of Embedded Hardware 4.2 Environment Configuration 4.3 System Operation Test and Analysis 5 Conclusion References IVOCT Image Based Coronary Artery Stent Detection Reconstruction 1 Introduction 2 Method 2.1 Median Filtering 2.2 Otsu Method 2.3 Canny Edge Detection Algorithm 3 Experimental Results and Discussion 3.1 Experimental Data 3.2 Image Preprocessing 3.3 Stent Identification 3.4 Coronary Stent Reconstruction 3.5 2D Reconstruction of Coronary Stents at Different Distances Between IVOCT Image Frames 4 Conclusion References Concept Drift Detection and Update Algorithm Based on Online Restricted Boltzmann Machine 1 Introduction 2 Basic Theory 2.1 Online Restricted Boltzmann Machine 2.2 Attention 3 O-RBM-Based Concept Drift Update Algorithm 4 Experiments and Analysis 4.1 Dataset 4.2 Experimental Results 5 Conclusion References 3D Data Augmented Person Re-identification and Edge-based Implementation 1 Introduction 2 3D Augmented Person Re-identification 3 Edge-based Implementation 4 Result and Analysis 5 Conclusion References 5G Potential Customer Recognition Research Based on Multi-layer Heterogeneous Integration Model 1 Introduction 2 5G Potential Customer Recognition Model 2.1 Problem Definition and Variable System Construction 2.2 Feature Extraction and Sample Balancing 2.3 Tree-Based Homogeneous Integrated Model 2.4 Deep Neural Network Model 2.5 Multi-layer Heterogeneous Integrated Model 3 Experimental Results and Analysis 3.1 5G Customer Feature Analysis and Sample Balancing Results 3.2 Different Classifier Experimental Results 3.3 Multi-layer Heterogeneous Integrated Model Experimental Results 4 Conclusion and Summary References Dynamic Task Offloading for Air-Terrestrial Integrated Networks: A Learning Approach 1 Introduction 2 System Model and Problem Formulation 2.1 Communication Model 2.2 Computing Model 2.3 Problem Formulation 3 Problem Solution 3.1 DRL-Based Offloading Decision Making 3.2 Greedy-Based Computing Resource Allocation 4 Performance Analysis and Simulations 5 Conclusion References Distributed Dynamic Spectrum Access for D2D Communications Underlying Cellular Networks Using Deep Reinforcement Learning 1 Introduction 2 System Model 2.1 Channel Model 2.2 Uplink Signal Model 3 DRL-based Dynamic Spectrum Access Scheme 3.1 Deep Reinforcement Learning 3.2 Uplink Dynamic Spectrum Access Framework 4 Performance Evaluation 5 Conclusion References Maximizing Energy-Efficiency in Wireless Communication Systems Based on Deep Learning 1 Introduction 2 System Model and Iterative Algorithm for EE Maximization 3 Deep Neural Network Based Method 3.1 DNN System Architecture 3.2 Data Generation and Pre-processing 3.3 Network Architecture 3.4 Training Stage 4 Results and Analysis 4.1 Simulation Environment 4.2 Results 4.3 Results Analyze 5 Conclusion References Dynamics Analysis and Optimisation of Television Jumpers’s Safety Clamp 1 Introduction 2 The Establishment of the Three-Dimensional Model of the Safety Clamp 3 ADAMS-Based Dynamics Simulation of the Safety Clamp 3.1 Calculation of the Braking Acceleration 4 Dynamics Analysis and Optimisation of Television Jumpers’s Safety Clamp 4.1 Calculation of Braking Distance 4.2 Analysis of Simulation Results 5 Safety Clamp Parameterised Design and Optimisation 5.1 GISTIFF Rigid Integration Algorithm 5.2 Optimisation Settings 5.3 Optimisation of Safety Clamp Braking Performance 6 Conclusion References Financial Time Series Forecast of Temporal Convolutional Network Based on Feature Extraction by Variational Mode Decomposition 1 Introduction 2 Methodology 2.1 Variational Mode Decomposition 2.2 Temporal Convolutional Network 2.3 Proposed Method 3 Experiment 3.1 Evaluation Metrics and Benchmark Experiments 3.2 Google Stock Experiments 3.3 GBP/CNY Exchange Rate Experiments 3.4 USD/CNY Exchange Rate Experiment 4 Summary References User Selection and Resource Allocation for Satellite-Based Multi-task Federated Learning System 1 Introduction 2 System Model 2.1 Task Model 2.2 Communication Model 2.3 Federated Learning Model 2.4 Optimization Problem 3 Optimization of Task Selection and Resource Allocation 4 Simulation Results 5 Conclusion References Industrial Time-Series Signal Anomaly Detection Based on G-LSTM-AE Model 1 Introduction 2 Anomaly Detection Based on G-LSTM-AE Model 2.1 LSTM-AE Network 2.2 Gaussian Confidence Interval 2.3 Detection Framework Based G-LSTM-AE 2.4 Industrial Time-Series Signal Anomaly Detection on G-LSTM-AE Model 3 Experiments and Dissuasions 3.1 Experimental Datasets 3.2 Experimental Settings 3.3 Experimental Results 4 Conclusion References Construction and Analysis of Scientific Research Knowledge Graph in the Field of Hydrogen Energy Technology 1 Introduction 2 Research Design 2.1 Knowledge Graph Framework Design 2.2 Data Source 3 Knowledge Graph Construction 3.1 Knowledge Graph Schema Layer Design 3.2 Knowledge Graph Data Layer Construction 4 Knowledge Graph Analysis 4.1 Research Status Analysis 4.2 Knowledge Association Analysis 5 Conclusions References Problems and Strategies of Foreign Language Education Development from the Perspective of Linguistic Intelligence 1 Introduction 2 Problems in the Development of Foreign Language Education from the Perspective of Linguistic Intelligence 2.1 The Development of Linguistic Intelligence Has Reduced the Demand for Foreign Language Professionals 2.2 Teachers Engaged in Foreign Language Education Generally Lack Language Intelligence Literacy 2.3 Teaching Methods Fail to Achieve Organic Integration with Language Intelligence Technology 3 Strategies for the Development of Foreign Language Education from the Perspective of Linguistic Intelligence 3.1 To Build a Compound Teaching Staff to Meet the Needs of Language Intelligence Development 3.2 To Build a Smart Foreign Language Education System 4 Conclusion References A Brief Analysis of AI-Empowered Foreign Language Education 1 Introduction 2 The Current Applications of the Language Intelligence Research 2.1 Intelligent Teaching System 2.2 Intelligent Evaluation System 2.3 Intelligent Corpus System 3 The Development Prospect of AI-Empowered Foreign Language Education in China 3.1 To Construct an Intelligent Teaching System of Foreign Language Education Based on the Research of Language Intelligence 3.2 To Truly Realize the Intelligentization of Foreign Language Education with Education Wisdom and Modern Technology 3.3 To Promote the Transformation of Research Results and Realize the Effective Docking of All Aspects of Foreign Language Education with Language Intelligent Technology 4 Conclusion References Complex-Valued Neural Networks with Application to Wireless Communication: A Review 1 Introduction 2 Why Use CVNNs? 3 CVNN Activation Functions and Structure 3.1 Complex Activation Functions 3.2 CVNN Structure 4 Wireless Radio CVNN Applications 4.1 Modulation Classification 4.2 Signal Recognition 4.3 Channel State Information (CSI) - Sensing and Prediction 4.4 Specific Emitter Identification (Aka RF Fingerprinting) 5 Future Research 5.1 Metaheuristic Learning Algorithms 5.2 Complex-Valued Activation Function Design References Author Index