International Conference on Neural Computing for Advanced Applications: 4th International Conference, NCAA 2023, Hefei, China, July 7–9, 2023, ... in Computer and Information Science, 1870)

✍ Scribed by Haijun Zhang (editor), Yinggen Ke (editor), Zhou Wu (editor), Tianyong Hao (editor), Zhao Zhang (editor), Weizhi Meng (editor), Yuanyuan Mu (editor)

• Publisher: Springer
• Year: 2023
• Tongue: English
• Leaves: 627
• Category: Library

✦ Synopsis

The two-volume set CCIS 1869 and 1870 constitutes the refereed proceedings of the 4th International Conference on Neural Computing for Advanced Applications, NCAA 2023, held in Hefei, China, in July 2023.
The 83 full papers and 1 short paper presented in these proceedings were carefully reviewed and selected from 211 submissions. The papers have been organized in the following topical sections: Neural network (NN) theory, NN-based control systems, neuro-system integration and engineering applications; Machine learning and deep learning for data mining and data-driven applications; Computational intelligence, nature-inspired optimizers, and their engineering applications; Deep learning-driven pattern recognition, computer vision and its industrial applications; Natural language processing, knowledge graphs, recommender systems, and their applications; Neural computing-based fault diagnosis and forecasting, prognostic management, and cyber-physical system security; Sequence learning for spreading dynamics, forecasting, and intelligent techniques against epidemic spreading (2); Applications of Data Mining, Machine Learning and Neural Computing in Language Studies; Computational intelligent Fault Diagnosis and Fault-Tolerant Control, and Their Engineering Applications; and Other Neural computing-related topics.

✦ Table of Contents

Preface
Organization
Contents – Part II
Contents – Part I
Deep Learning-Driven Pattern Recognition, Computer Vision and Its Industrial Applications
Improved YOLOv5s Based Steel Leaf Spring Identification
1 Introduction
2 YOLOv5 Structure and Method Flow
2.1 Steel Leaf Spring Visual Identification Process
2.2 YOLOv5s Network Structure
3 YOLOv5 Recognition Algorithm Improvement
3.1 YOLOv5 Steel Leaf Spring Recognition Based On Migration Learning
3.2 CBAM Convolutional Attention Mechanism
3.3 Network Model Lightweighting
4 Experimental Results and Analysis.
4.1 Ablation Experiments
4.2 Comprehensive Comparison Experiments of Different Target Detection Models
5 Summary
References
A Bughole Detection Approach for Fair-Faced Concrete Based on Improved YOLOv5
1 Introduction
2 Model Design
2.1 The Network Structure of YOLOv5
2.2 Network Structure Improvement
3 Experimental Settings and Results
3.1 The Experiment Platform
3.2 Data Acquisition and Dataset
3.3 Evaluation Metrics
3.4 Experimental Results and Analysis
4 Conclusion
References
UWYOLOX: An Underwater Object Detection Framework Based on Image Enhancement and Semi-supervised Learning
1 Introduction
2 UWYOLOX
2.1 Joint Learning-Based Image Enhancement Module (JLUIE)
2.2 Improved Semi-supervised Learning Method for Underwater Object Detection (USTAC)
3 Experiments
3.1 Implementation Details
3.2 Experiment Results
4 Discussion and Conclusion
References
A Lightweight Sensor Fusion for Neural Visual Inertial Odometry
1 Introduction
2 Relate Work
2.1 VO
2.2 Traditional VIO Methods
2.3 Deep Learning-Based VIO
3 Method
3.1 Attention Mechanism for the Visual Branch
3.2 Lightweight Pose Estimation Module
3.3 Loss Function
4 Experiment
4.1 Dataset
4.2 Experimental Setup and Details
4.3 Main Result
5 Conclusion
References
A Two-Stage Framework for Kidney Segmentation in Ultrasound Images
1 Introdution
2 Relate Works
2.1 Automated Kidney Ultrasound Segmentation
2.2 Level-Set Function
2.3 Self-correction
3 Method
3.1 Overview
3.2 Shape Aware Dual-Task Multi-scale Fusion Network
3.3 Self-correction Part
4 Experiments
4.1 Dataset and Implementation Details
4.2 Experiment Results
4.3 Ablation Studies
5 Conclusion
References
Applicability Method for Identification of Power Inspection Evidence in Multiple Business Scenarios
1 Introduction
2 Constructing a Sample Library for Identifying Power Inspection Supporting Materials
3 Text Recognition Based on YOLOv3 Network
4 Network Compression with Structure Design and Knowledge Distillation
5 Experiment and Analysis
5.1 Training Sample Augmentation Quality Assessment
5.2 Model Recognition Results and Analysis
5.3 Application Effect of Intelligent Verification in Power Inspection
6 Conclusion
References
A Deep Learning Algorithm for Synthesizing Magnetic Resonance Images from Spine Computed Tomography Images using Mixed Loss Functions
1 Introduction
2 Related Work
3 Method
3.1 Generator
3.2 Discriminator
3.3 Mixed loss function
4 Experiment and Analysis
4.1 Paired Dataset and Pretreatment
4.2 Experimental Setup and Evaluation Methods
4.3 Experimental Result
5 Conclusion
References
Investigating the Transferability of YOLOv5-Based Water Surface Object Detection Model in Maritime Applications
1 Introduction
2 Related Work
2.1 Object Detection
2.2 Yolov5
2.3 Public Dataset
3 Method
3.1 Dataset
3.2 Metric
4 Experiments
5 Conclusion
References
Physical-Property Guided End-to-End Interactive Image Dehazing Network
1 Introduction
2 Related Work
2.1 Physics-Based Dehazing Methods
2.2 End-to-End Deep Dehazing Methods
3 Proposed Method
3.1 End-to-End Information Interaction Network
3.2 Color-Detail Refinement Sub-Network (R-Net)
3.3 Loss Function
4 Experiments
4.1 Experimental Settings
4.2 Quantitative Dehazing Results
4.3 Visual Image Analysis
4.4 Ablation Studies
5 Conclusion
References
A Clothing Classification Network with Manifold Structure Based on Second-Order Convolution
1 Introduction
2 Related Work
2.1 Clothing Classification
2.2 Manifold Structured Neural Network
3 Method
3.1 Covariance Pooling
3.2 Second Order Convolution on SPD Manifolds
3.3 Riemann Batch Normalization of SPD Matrices
3.4 Rectified Linear Units
3.5 Parameter Vectorization Layer
4 Experiments
4.1 Dataset
4.2 Details
4.3 Results
5 Conclusion
References
Multi-size Scaled CAM for More Accurate Visual Interpretation of CNNs
1 Introduction
2 Related Work
2.1 Backpropagation Based Saliency Methods
2.2 Perturbation-Based Saliency Methods
3 Proposed Approach
3.1 Grad-CAM
3.2 Masks Generation
3.3 Masks Optimization
3.4 Saliency Map Generation
4 Experiments
4.1 Experiments Setup
4.2 Qualitative Evaluation
4.3 Insertion and Deletion
4.4 Localization Evaluation
4.5 Sanity Check
5 Conclusion
References
Joint Attention Mechanism of YOLOv5s for Coke Oven Smoke and Fire Recognition Algorithm
1 Introduction
2 The Main Problems with Coke Oven Smoke and Fire Detection
3 Design of a Coke Oven Smoke and Fire Detection Model
3.1 YOLOv5s Model
3.2 Attention Mechanisms
3.3 YOLOv5s for Joint Attention Mechanisms
4 Experiments and Analysis of Results
4.1 Constructing the Dataset
4.2 Model Training
4.3 Comparative Analysis of Experimental Results
5 Conclusion
References
Natural Language Processing, Knowledge Graphs, Recommender Systems, and Their Applications
An Enhanced Model Based on Recurrent Convolutional Neural Network for Predicting the Stage of Chronic Obstructive Pulmonary Diseases
1 Introduction
2 Methods
2.1 Enhanced Recurrent Convolutional Neural Network
2.2 Pre-training Word Embedding
2.3 Evaluation Metrics
3 Experiments and Results
3.1 Dataset
3.2 Baseline Methods
3.3 Parameters Setting
3.4 Descriptive Analysis
4 Conclusions
References
Hybrid Recommendation System with Graph Neural Collaborative Filtering and Local Self-attention Mechanism
1 Introduction
2 Preliminaries
2.1 NGCF
2.2 Local Self-attention
3 Proposed Model
3.1 Data Preprocessing
3.2 Embedding Layer
3.3 Embedding Propagetion Layer
3.4 Prediction Layer
3.5 Evaluate
4 Experiments
4.1 Evaluation Metrics
4.2 Results
4.3 Analyse
5 Conclusion
References
MAMF: A Multi-Level Attention-Based Multimodal Fusion Model for Medical Visual Question Answering
1 Introduction
2 Related Work
3 Methods
3.1 Problem Formulation
3.2 Overview of Our Proposed Model
3.3 Word Embedding and Question Representation
3.4 Visual Feature Extractor
3.5 Attention-Based Multimodal Fusion Module
3.6 Loss Function
4 Experiments
4.1 Datasets
4.2 Experiment Settings
4.3 Baseline Models
4.4 Results
4.5 Ablation Study
4.6 Hyperparameter Analysis
4.7 Qualitative Evaluation
5 Conclusion
References
ASIM: Explicit Slot-Intent Mapping with Attention for Joint Multi-intent Detection and Slot Filling-1pc
1 Introduction
2 Related Works
2.1 Intent Detection Tasks
2.2 Slot Filling Tasks
2.3 Joint Model
3 Methodology
3.1 Problem Definition
3.2 ASIM Model
3.3 Encoder
3.4 Decoder
3.5 Asynchronous Training
4 Experimental Results
4.1 The Data-Set Description
4.2 Baselines
4.3 The Experiment Design
4.4 The Experiment Results
5 Conclusions
References
A Triplet-Contrastive Representation Learning Strategy for Open Intent Detection-1pc
1 Introduction
2 Related Work
3 Method
3.1 Definition
3.2 The Architecture of TCAB
3.3 Feature Extraction
3.4 Triplet-Contrastive Representation Learning
3.5 Open Intent Detection
4 Experiments
4.1 Datasets and Evaluation Metrics
4.2 Baselines and Settings
4.3 Main Results and Ablation Study
4.4 Analysis
5 Conclusions
References
A User Intent Recognition Model for Medical Queries Based on Attentional Interaction and Focal Loss Boost
1 Introduction
2 Related Work
3 The TAI-FLB Model
3.1 Query Sentence Embedding and Label Embedding
3.2 Attentional Interaction
3.3 Focal Loss Boost
4 Experiments and Results
4.1 Datasets
4.2 Evaluation Metrics
4.3 Parameter Settings
4.4 Baseline Methods
4.5 The Result
5 Conclusions
References
Neural Computing-Based Fault Diagnosis and Forecasting, Prognostic Management, and Cyber-Physical System Security
Multiscale Redundant Second Generation Wavelet Kernel-Driven Convolutional Neural Network for Rolling Bearing Fault Diagnosis
1 Introduction
2 Theoretical Foundation
2.1 RSGW Transform
2.2 Fundamental Theory of CNN
3 The Proposed Method
3.1 Design of RSGW Convolution Kernel
3.2 Deep CNN Driven by Multiscale RSGW Kernels
4 Experimental Verification
4.1 Case1: CWRU
4.2 Case2: JNU
5 Conclusion
References
Unsupervised Deep Transfer Learning Model for Tool Wear States Recognition
1 Introduction
2 Background and Preliminaries
2.1 Data Distribution Adaptation of Deep Transfer Learning
2.2 Data Distribution Alignment
3 Proposed Methodology
3.1 Monitoring Data Processing Method
3.2 Tool Wear States Recognition Neural Network
3.3 Model Learning Task
3.4 Model Training Method
4 Experiment and Discussion
4.1 Milling Experiment Setup
4.2 Data Preprocessing
4.3 Model Performance Analysis
5 Conclusion
References
Fault Diagnosis of High-Voltage Circuit Breakers via Hybrid Classifier by DS Evidence Fusion Algorithm
1 Introduction
2 Technical Framework
3 Theoretical Background
3.1 Variational Mode Decomposition
3.2 DS Evidence Fusion Algorithm
4 Experimental Application
4.1 Experiment Setup
4.2 Feature Extraction
5 Result Analysis
6 Conclusion
References
Multi-Feature Fusion and Reinforcement Model for High-Speed Train Axle Box Bearing Fault Diagnosis Under Variable Speed Domain
1 Introduction
2 Method
2.1 Multi-Scale Feature Extraction Network
2.2 Multi-Scale Feature Fusion Mechanism
2.3 Feature Reinforcement Mechanism
3 Experiment Verification
3.1 Bogie Axlebox Bearing Fault Simulation Test Bench
3.2 Classification Comparison and Analysis
4 Conclusion
References
Degradation Modelling and Remaining Useful Life Prediction Methods Based on Time Series Generative Prediction Networks
1 Introduction
2 The Proposed Degradation Trend Prediction Model
2.1 GAN Method
2.2 GRU Network
2.3 The GAN-GRU Network
3 Experimental Verification and Analysis
3.1 Using GAN to Generate Data
3.2 Prediction Results of Bearing Data Set PHM-2012
4 Conclusion
References
Sequence Learning for Spreading Dynamics, Forecasting, and Intelligent Techniques Against Epidemic Spreading (2)
Data-Model Intergrowth Makes Better Time Series Prediction
1 Introduction
2 Methodology
2.1 Problem Modeling
2.2 Initialization
2.3 Iteration Process
3 Experiment
3.1 Datasets
3.2 Results
3.3 Discussion
4 Conclusion
References
Spatial-Temporal Electric Vehicle Charging Demand Forecasting: A GTrans Approach-1pc
1 Introduction
2 Methodology
2.1 Problem Statement
2.2 Modeling
2.3 Loss Function
3 Experiments and Data
3.1 Data
3.2 Evaluation Function
3.3 Model Settings
3.4 Results and Analysis
4 Conclusion
References
A Long Short-term Memory Model for COVID-19 Forecasting Using High-efficiency Feature Representation
1 Introduction
2 Related Works
2.1 Empirical Wavelet Transform
2.2 Variational Mode Decomposition
2.3 Long Short-Term Memory
3 Methodologies
3.1 Problem Formulation
3.2 Data Augmentation
3.3 Proposed EWT-VMD-LSTM Model
4 Experiments and Results
4.1 Dataset and Preprocessing
4.2 Experiment Settings and Evaluation Metrics
4.3 Forecasting Performance
4.4 Comparisons and Ablation Study
5 Conclusion
References
Stepwise Fusion Transformer for Affective Video Content Analysis
1 Introduction
2 Related Work
2.1 Transformer
2.2 Affective Video Content Analysis
3 Proposed Method
3.1 Features
3.2 Model
3.3 Loss Functions
4 Experiments
4.1 Datasets
4.2 Implementation Details
4.3 Experimental Results
4.4 Ablation Study
5 Conclusion
References
An Adaptive Clustering Approach for Efficient Data Dissemination in IoV
1 Introduction
2 Related Work
2.1 Routing Algorithms in VANET
2.2 Cluster Algorithms in VANET
3 Clustering Management Scheme
3.1 System Model
3.2 Clustering Management Strategy
4 Bp-ADOV Routing Algorithm
4.1 Reference Factors for Routing Algorithm
4.2 Adaptive Clustering Routing Algorithm Combined with Reinforcement Learning
4.3 Message Propagation Scheme
5 Simulations
5.1 Bp-AODV Simulation Settings
5.2 Analysis of Experimental Results
6 Conclusion
References
Strategy Determination for Multiple USVs: A Min-max Q-learning Approach
1 Introduction
2 Preliminaries and Problem Statement
2.1 Motions for the USV
2.2 Problem Formulation
3 The Specific Min-max Q-learning
3.1 Construction of the Learning State Space
3.2 Calculation of the Q-function
3.3 Reward Function
4 Simulations
4.1 Case 1
4.2 Case 2
5 Conclusions
References
COVID-19 Urban Emergency Logistics Planning with Multi-objective Optimization Model
1 Literature Review
1.1 Traditional Emergency Logistics
1.2 Recent Studies Focusing on Emergency Logistics Under Pandemic
1.3 Multi-Objective Optimization Problem
2 Model Establishment
2.1 Weight Calculation
2.2 Application of Efficacy Coefficient Method
3 Optimal Vehicle Routing Planning
3.1 Problem Statement
3.2 Vehicle Routing Planning Based on Non-Dominated Sorting Genetic Algorithm
4 Simulation and Scheme of Route Plan
4.1 Data Analyses
4.2 Results Analysis and Comparison
5 Conclusions and Future Work
References
Joint Data Routing and Service Migration via Evolutionary Multitasking Optimization in Vehicular Networks
1 Introduction
2 System Architecture
3 Problem Formulation
3.1 Preliminary
3.2 Data Routing Problem
3.3 Service Migration Problem
4 Proposed Solution
4.1 Evolutionary Solver Setting
4.2 Knowledge Transfer Strategy
5 Experiment
5.1 Set up
5.2 Simulation Results
6 Conclusion
References
Applications of Data Mining, Machine Learning and Neural Computing in Language Studies
Teaching Pre-editing for Chinese-to-English MT: An Experiment with Controlled Chinese Rules
1 Introduction
2 Pre-editing and Controlled Chinese Rules
2.1 Literature Review on Pre-editing and Controlled Language
2.2 Controlled Chinese Rules for Web-based Business Text
3 The Teaching Experiment
3.1 Participants
3.2 Procedure
4 Data Analysis and Discussion
4.1 Analysis on Students’ Application of Rule 1—Every Sentence Should Have an Explicit Subject
4.2 Analysis on Students’ Application of Rule 2—Sentences Should be Short
4.3 Analysis on Students’ Application of Rule 3—There Should be no Repetitive Complimentary Expressions
4.4 Students’ Feedback
5 Conclusion
References
Research on the Application of Computer Aided Corrective Feedback in Foreign Language Grammar Teaching
1 Introduction
2 Literature Review
3 Research Design
3.1 Research Topics
3.2 Research Participants
3.3 Test Materials
3.4 Research Methods
3.5 Data Analysis
4 Results
5 Discussion
6 Conclusion
References
Student-centered Education in Metaverse: Transforming the Language Listening Curriculum
1 Introduction
2 Theoretical Literature Review
3 Data Analysis
4 Instructional Methods
4.1 Teaching Resources
4.2 Teacher Education and Course Management
5 Student Perceptions
5.1 Student Literacy
5.2 Student Motivation
5.3 Student Technology Acceptance
6 Schema Knowledge via Metaverse
7 Conclusions
References
A Positive-Negative Dual-View Model for Knowledge Tracing
1 Introduction
2 Related Work
3 Preliminary
3.1 Problem Definition
3.2 Heterogeneous Information Network
4 The Approach
4.1 Input Representation
4.2 Dual-View Model
4.3 Information Filtering Module
4.4 Student State Evolution
4.5 History Performance Interaction Module
4.6 Optimization
5 Experiments and Results
5.1 Datasets
5.2 Baselines
5.3 Implementation Details
5.4 Results
6 Conclusions
References
A Study of Chinese-English Translation Teaching Based on Data Mining
1 Introduction
2 Literature Review
3 Textwells Platform
4 The Relevance Visualization of Translation Knowledge Nodes
4.1 Gephi-a Network Visualization Software
4.2 Translation Knowledge Nodes Relevance Network Visualization
4.3 Translation Knowledge Nodes Relevance Analysis
5 Reflections of the Teaching Design of the Current Translation Courses
6 Conclusion
References
Computational Intelligent Fault Diagnosis and Fault-Tolerant Control, and Their Engineering Applications
Novel TD-based Adaptive Control for Nonlinearly Parameterized Stochastic Systems
1 Introduction
2 Mathematical Preliminaries
2.1 System Description
2.2 Preliminaries
2.3 Main Controller Design Process
3 Simulation Example
4 Conclusion
References
A Data-driven Intermediate Estimator-based Approach for Collaborative Fault-tolerant Tracking Control of Multi-agent Systems
1 Introduction
2 Preliminaries and Problem Statement
2.1 Graph Theory
2.2 System Description
3 Design of Data-driven Distributed Intermediate Estimator
4 Convergence Analysis
5 Experimental Results
6 Conclusion
References
Anomaly Detection and Alarm Limit Design for In-Hole Bit Bounce Based on Interval Augmented Mahalanobis Distance
1 Introduction
2 Problem Description
3 The Proposed Method
3.1 Data Augmentation
3.2 Design Abnormal Monitoring Indicator
3.3 Alarm Limit Design
3.4 Procedures and Evaluation
4 Industrial Case Study
5 Conclusion
References
Other Neural Computing-Related Topics
A Neural Approach Towards Real-time Management for Integrated Energy System Incorporating Carbon Trading and Electrical Vehicle Scheduling-1pc
1 Introduction
2 Problem Statement
2.1 System Structure
2.2 Decision Problem Formulation
3 A Neural Approach to Real-time IES Management
4 Case Study
4.1 System Description
4.2 Optimal Scheduling Results Considering Carbon Trading
4.3 Optimal Scheduling Results Considering EVs and Carbon Trading
References
Research on Chinese Diabetes Question Classification with the Integration of Different BERT Models
1 Introduction
1.1 A Text Classification System Based on Machine Learning Algorithms
1.2 A Text Classification System Based on Deep Learning Algorithms
1.3 Fine Tuning of Text Classification Tasks Based on Language Pre-training Model
2 Related Work
2.1 Text Classification Tasks in the Medical Field
2.2 The Application of BERT Model in Classification Tasks
2.3 Different Models Based on BERT
3 Design Method of the Model
3.1 Brief Observation of Data
3.2 Normalize the Input Chinese Text
3.3 Using Pre-trained BERT Models as the Basis for Input
3.4 Building a Convolutional Neural Network
3.5 The Training Process of the Model
4 Experiment
4.1 Experiment Data
4.2 Evaluation Criterion
4.3 Experimental Results and Analysis
5 Conclusion
References
Shared Task 1 on NCAA 2023: Chinese Diabetes Question Classification
1 Introduction
2 Datasets
2.1 Distribution of the Dataset
2.2 Evaluation
2.3 Baseline
3 The Results
4 Conclusion
References
SFDA: Chinese Diabetic Text Classification Based on Sentence Feature Level Data Augmentation
1 Introduction
2 Related Work
3 Approach
3.1 Semantic Feature Enhancing Module
3.2 Serialized Attention Aggregation Module
4 Experiment
4.1 Dataset
4.2 Experimental Setting
4.3 Evaluation Metric
4.4 Experimental Results and Analysis
5 Conclusion
References
Author Index