Neural Information Processing: 30th International Conference, ICONIP 2023, Changsha, China, November 20–23, 2023, Proceedings, Part XI (Communications in Computer and Information Science)

✍ Scribed by Biao Luo (editor), Long Cheng (editor), Zheng-Guang Wu (editor), Hongyi Li (editor), Chaojie Li (editor)

• Publisher: Springer
• Year: 2023
• Tongue: English
• Leaves: 617
• Category: Library

✦ Synopsis

The nine-volume set constitutes the refereed proceedings of the 30th International Conference on Neural Information Processing, ICONIP 2023, held in Changsha, China, in November 2023.
The 1274 papers presented in the proceedings set were carefully reviewed and selected from 652 submissions.
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.

✦ Table of Contents

Preface
Organization
Contents – Part XI
Applications
Multi-intent Description of Keyword Expansion for Code Search
1 Introduction
2 Model Method
2.1 Training Model Selection
2.2 Joint Embedding
2.3 Extended Research
3 Experimental Analysis
3.1 Experimental Preparation
3.2 Difference Analysis
3.3 Comparative Experimental Analysis
3.4 Analysis of Ablation Experiments
4 Conclusion
References
Few-Shot NER in Marine Ecology Using Deep Learning
1 Introduction
2 Few-Shot NER Based on Deep Learning
2.1 Text Generation Using SeqGAN
2.2 Distributed Representations for Input Using BERT
2.3 Context Encoder Using IDCNN
2.4 Context Encoder Using BiLSTM
2.5 Tag Decoder Using CRF
3 Data Processing and Evaluation Metrics
3.1 Data Preprocessing
3.2 Labeling Guidelines
3.3 Evaluation Metrics
4 Results and Analysis
5 Conclusion
References
Knowledge Prompting with Contrastive Learning for Unsupervised CommonsenseQA
1 Introduction
2 Related Works
3 Methodology
3.1 Further Pre-training with Contrastive Learning
3.2 Knowledge Generation with Generic Prompt
3.3 Knowledge Reasoning and Answer Prediction
4 Experiments and Results
4.1 Datasets and Metric
4.2 Baselines and Implementation Details
4.3 Main Results
4.4 Ablation Study
4.5 Robustness Analysis
4.6 Case Study
5 Conclusion
References
PTCP: Alleviate Layer Collapse in Pruning at Initialization via Parameter Threshold Compensation and Preservation
1 Introduction
2 Related Work
3 Method
3.1 Problem Formulation
3.2 Parameter Threshold Compensation
3.3 Parameter Preservation
3.4 Round-by-Round Matching Algorithm
4 Experiment
4.1 Experimental Settings
4.2 Results and Analysis
4.3 Ablation Study
5 Conclusions
References
Hierarchical Attribute-Based Encryption Scheme Supporting Computing Outsourcing and Time-Limited Access in Edge Computing
1 Introduction
2 Preliminaries
2.1 Bilinear Groups Pairing
2.2 Linear Secret Sharing Scheme
2.3 Security Assumption
3 System Structure
3.1 System Model
3.2 Scheme Description
4 Security Proof
5 Performance Analysis
5.1 Security Analysis
5.2 Performance Analysis
6 Conclusion
References
An Ontology for Industrial Intelligent Model Library and Its Distributed Computing Application
1 Introduction
2 Related Work
2.1 Ontology for Intelligent Manufacturing
2.2 Open Challenges
3 The Development of the Ontology
3.1 The Schema Layer of the Ontology
3.2 The Data Layer of the Ontology
4 Application of the Ontology
4.1 Actors
4.2 Distribute Computing Working Flow
5 Conclusion
References
Efficient Prompt Tuning for Vision and Language Models
1 Introduction
2 Related Work
2.1 Single-modal Prompt Tuning
2.2 Multi-modal Prompt Tuning
3 Approach
3.1 Visual and Language Pre-training
3.2 Visual Prompt Tuning
3.3 Text Prompt Tuning
3.4 Downstream Task-Related Loss Fusion
4 Experiments and Discussions
4.1 Few-Shot Learning
4.2 Performance of the Model with Different Loss Functions
5 Conclusion
References
Spatiotemporal Particulate Matter Pollution Prediction Using Cloud-Edge Intelligence
1 Introduction
2 Proposed PM2.5 Prediction Model
2.1 Data Collection and Preprocessing
2.2 CNN-LSTM Model Architecture
2.3 Cloud-Edge Model Training and Inference
3 Experimental Work
3.1 Setup and Dataset
3.2 Results and Discussions
4 Conclusion and Future Work
References
From Incompleteness to Unity: A Framework for Multi-view Clustering with Missing Values
1 Introduction
2 Related Work
3 Methodology
3.1 Distance Estimation
3.2 Distance and Affinity Correction
3.3 Theoretical Analysis
4 Experiments
4.1 Experimental Setup
4.2 Incomplete Clustering on Single-View Data
4.3 Incomplete Clustering on Multi-view Data
4.4 Quality Visualization of Affinity Matrices
4.5 Motivation and Results Summary
5 Conclusion and Future Work
References
PLKA-MVSNet: Parallel Multi-view Stereo with Large Kernel Convolution Attention
1 Introduction
2 Related Work
3 Proposed Method
3.1 Network Overview
3.2 Feature Extractor
3.3 Cost Volume Construction
3.4 Parallel Cost Volume Aggregation (PCVA)
3.5 Loss Function
4 Experiments
4.1 Datasets
4.2 Implementation
4.3 Experimental Results
4.4 Ablation Study
5 Conclusion
References
Enhancement of Masked Expression Recognition Inference via Fusion Segmentation and Classifier
1 Introduction
2 Related Work
2.1 Subregion-Based Methods
2.2 Reconstruction-Based Methods
2.3 Discard-Based Methods
3 Proposed Method
3.1 Occlusion Awareness Module
3.2 Occlusion Purification Module
3.3 Expression Information Capture Module
4 Experiments
4.1 Datasets
4.2 Implementation Details
4.3 Results Comparison
4.4 Ablation Study
4.5 Visualization
5 Conclusion
References
Semantic Line Detection Using Deep-Hough with Attention Mechanism and Strip Convolution
1 Introduction
2 Related Work
2.1 Hough Transform
2.2 CNN-Based Line Detection
2.3 Semantic Lines
2.4 Semantic Lines Detection
3 Proposed Method
3.1 Network Structure
3.2 Feature Selection Module
3.3 Strip Convolution
3.4 Mixed Strip Pooling Layer
3.5 Lightweight Network Using the Strategy of GhostNet
4 Experiments
4.1 Comparison of Experimental Results
4.2 Ablation Study
5 Conclusion
References
Adaptive Multi-hop Neighbor Selection for Few-Shot Knowledge Graph Completion
1 Introduction
2 Related Work
3 Preliminaries
4 Methodology
4.1 Neighbor Aggregator
4.2 Transformer Encoder
4.3 Attentional Matching Network
4.4 Loss Function
5 Experiments
5.1 Datasets and Baselines
5.2 Implementation
5.3 Experimental Comparison with Baselines
5.4 Comparison over Different Relations
5.5 Ablation Study
6 Conclusion
References
Applications of Quantum Embedding in Computer Vision
1 Introduction
2 Related Works
2.1 Revisiting QSM
2.2 Attention Mechanism in CNNs
2.3 Self-organizing Map
3 Proposed Method
3.1 QE-SOM
3.2 Quantum-State-Based Attention Networks
4 Experiments
4.1 Experiments on MNIST
4.2 Experiments on ImageNet
5 Conclusion
References
Traffic Accident Forecasting Based on a GrDBN-GPR Model with Integrated Road Features
1 Introduction
2 Methodology
2.1 Gaussian Radial Deep Belief Network
2.2 Gaussian Process Regression
2.3 Proposed Method
3 Case Study
3.1 Data Description and Preprocessing
3.2 Experiment 1
3.3 Experiment 2
4 Conclusions
References
Phishing Scam Detection for Ethereum Based on Community Enhanced Graph Convolutional Networks
1 Introduction
2 Related Works
2.1 Phishing Detection Methods
2.2 Network Representation Learning
3 Proposed Methods
3.1 Problem Statement
3.2 Feature Extraction
3.3 Community Enhanced Phishing Scam Detection
4 Experiment
4.1 Datasets
4.2 Evaluation Metrics
4.3 Comparison with Other Methods
4.4 Evaluation of Model Parameters
5 Conclusion and Future Work
References
DTP: An Open-Domain Text Relation Extraction Method
1 Introduction
2 Methodology
2.1 Encoder
2.2 Open Relation Detection
2.3 Open Relation Discovery
3 Experiments
3.1 Baselines and Evaluation Metrics
3.2 Parameters Setting and Training Details
3.3 Result and Discussion
4 Conclusions and Future Works
References
Exploring the Capability of ChatGPT for Cross-Linguistic Agricultural Document Classification: Investigation and Evaluation
1 Introduction
2 Related Work
2.1 Agricultural Document Classification
2.2 ChatGPT
3 ChatGPT-Based Agricultural Document Classification
3.1 Prompt Question Construction
3.2 ChatGPT Q&A Inference
3.3 Answer Alignment
4 Experiments
4.1 Setups
4.2 Main Results
4.3 Improved with ChatGPT Triggered Prompts
4.4 Improved with PromptPerfect Prompt Optimization
5 Prominent Cross-Linguistic Supports
6 Conclusion
References
Multi-Task Feature Self-Distillation for Semi-Supervised Machine Translation
1 Introduction
2 Method
2.1 Teacher Model Structure
2.2 Working Specification
3 Experiments
3.1 Datasets
3.2 Baselines
3.3 Evaluation
3.4 Implementation Details
4 Experimental Results
4.1 Results of Standard Datasets
4.2 Results of Low Resource Datasets
4.3 Results of Domains Datasets
4.4 Domain Adaptation
5 Analysis
5.1 Effect of Language Modeling Task Training Objectives
5.2 Effect on Different Part
5.3 Monolingual Data Size
5.4 Hallucinations
5.5 Computation Overhead
6 Conclusion
References
ADGCN: A Weakly Supervised Framework for Anomaly Detection in Social Networks
1 Introduction
2 Related Work
2.1 Anomaly Detection in Social Network
2.2 Graph Unsupervised Learning
2.3 Message Passing Neural Networks
2.4 Weakly Supervised Learning
3 Methodology
3.1 Information-Preserving Feature Compression
3.2 Collaborative Mining of Global and Local Information
3.3 Multi-view Weakly Supervised Classifier
4 Experiment
4.1 Dataset
4.2 Baselines (RQ1)
4.3 Ablation Experiment (RQ2)
4.4 Evaluation of Node Representation (RQ3)
5 Discussion
5.1 Hyperparameter Analysis
5.2 Loss Convergence
6 Conclusion
References
Light Field Image Super-Resolution via Global-View Information Adaption and Angular Attention Fusion
1 Introduction
2 Architecture and Pipeline
2.1 Overview
2.2 Spatial-Angular Feature Extraction Module (S-AFEM)
2.3 Global-View Adaptation-Guided Module (GAGM)
2.4 Angular Attention Fusion Module (AAFM)
2.5 Fusion and Upsampling Module (FM)
3 Experiments
3.1 LF Public Datasets and Evaluation Metrics
3.2 Settings and Implementation Details
3.3 Comparison to State-of-the-Art Methods
3.4 Ablation Study
4 Conclusion
References
Contrastive Learning Augmented Graph Auto-Encoder
1 Introduction
2 Related Work
2.1 Graph Embedding
2.2 Graph Constrative Embedding
3 The Proposed Method
3.1 Preliminary Work
3.2 Variational Graph AutoEncoder
3.3 Distribution-Dependent Regularization
3.4 Truncated Triplet Loss
4 Experiments
4.1 Datasets
4.2 Baselines and Implementation Details
4.3 Link Prediction
4.4 Node Clustering
4.5 Ablation Experiments
5 Conclusion
References
Enhancing Spatial Consistency and Class-Level Diversity for Segmenting Fine-Grained Objects
1 Introduction
2 Related Work
3 Method
3.1 Overall Network Architecture
3.2 Spatial Consistency Enhancement Module
3.3 Fine-Grained Regions Contrastive Loss
3.4 Loss
4 Experiments
4.1 Implementation Details
4.2 Results
4.3 Ablation Study
4.4 Visualization
5 Conclusion
References
Diachronic Named Entity Disambiguation for Ancient Chinese Historical Records
1 Introduction
2 Related Works
3 Method
3.1 Retriever
3.2 Reranker
3.3 Sample Re-Weighting by Frequency
3.4 Coreference Resolution Post-processing
4 Dataset
5 Experiments
5.1 Settings
5.2 Baselines and Results
5.3 Analysis
6 Discussion
7 Conclusion
References
Construction and Prediction of a Dynamic Multi-relationship Bipartite Network
1 Introduction
2 Related Work
3 DMBN Model
3.1 Framework Overview
3.2 DMBN Construction
3.3 Dynamic Prediction
4 Experiment and Results
4.1 Dataset
4.2 Evaluation Metrics
4.3 Baseline Methods
4.4 Results
5 Conclusion
References
Category-Wise Fine-Tuning for Image Multi-label Classification with Partial Labels
1 Introduction
2 Related Work
3 Methods
3.1 Category-Wise Fine-Tuning (CFT)
3.2 Greedy Selection for Fine-Tuning Configuration Selection
3.3 Fine-Tuning Logistic Regressions (LRs) Using Genetic Algorithm
4 Experimental Results and Discussion
4.1 The CheXpert Chest X-Ray Image MLC Competition Dataset
4.2 Partially Labeled Versions of MS-COCO
5 Conclusion
References
DTSRN: Dynamic Temporal Spatial Relation Network for Stock Ranking Recommendation
1 Introduction
2 Related Work
2.1 Stock Ranking Recommendation with Temporal-Spatial Relations
2.2 Stock Ranking Recommendation with Dynamic Graph
3 Preliminary
3.1 Stock Series Data
3.2 Stock Ranking Recommendation
4 Methodology
4.1 Temporal Relation Extraction
4.2 Dynamic Global-View Spatial Relation Extraction
4.3 Dynamic Multi-view Spatial Relation Extraction
4.4 Dynamic Temporal Spatial Relation Aggregation
4.5 Ranking Recommendation
5 Experiments
5.1 Datasets
5.2 Experimental Settings
5.3 Baselines Methods
6 Results and Analysis
6.1 Overall Performance
6.2 Ablation Study
7 Conclusion
References
Semantic Segmentation of Multispectral Remote Sensing Images with Class Imbalance Using Contrastive Learning
1 Introduction
2 Methodology
2.1 Semantic Consistency Constraint
2.2 Rebalancing Sampling Strategy
2.3 Pixel-Level Supervised Contrastive Loss
3 Experiments
3.1 Datasets
3.2 Experimental Setup
3.3 Experimental Results and Analysis
3.4 Parameter Analysis on the Number of Sampled Samples
3.5 Ablation Experiments
4 Conclusion
References
ESTNet: Efficient Spatio-Temporal Network for Industrial Smoke Detection
1 Introduction
2 Our Method
2.1 Shallow Enhanced Feature Extraction Module
2.2 Smoke Spatio-Temporal Feature Learning
2.3 Multi-temporal Spans Fusion Module
3 Experiments
3.1 Datasets
3.2 Implementation Details
3.3 Comparisons on RISE
3.4 Ablation Study
3.5 Visualization Analysis
4 Conclusion
References
Algorithm for Generating Tire Defect Images Based on RS-GAN
1 Introduction
2 Related Work
2.1 Attention Mechanism
2.2 Residual Networks and RSNet
3 RS-GAN Tire Defect Image Generative Model
3.1 RS-GAN Model Framework
3.2 Generators and Discriminators for RS-GAN
3.3 Optimization of Loss Function
4 Experimental Results and Analysis
4.1 Evaluation Indicators and Generated Image Display
4.2 Results and Analysis of Ablation Experiments
4.3 Results and Analysis of Comparative Experiments
5 Conclusion
References
Novel-Registrable Weights and Region-Level Contrastive Learning for Incremental Few-shot Object Detection
1 Introduction
2 Related Works
3 Method
3.1 Problem Formulation
3.2 Novel-Registrable Weights
3.3 Region-Level Contrastive Learning
4 Experiments
4.1 Experimental Setting
4.2 Incremental Few-Shot Object Detection
4.3 Continuous iFSOD
4.4 Ablation Study
5 Conclusion
References
Hybrid U-Net: Instrument Semantic Segmentation in RMIS
1 Introduction
2 Related Work
3 Method
3.1 Encoder
3.2 Modified Feature Pyramid Module: TSPP
3.3 Seamless Skip-Connection
3.4 Decoder
4 Experiments and Results
4.1 Datasets
4.2 Evaluation Metrics
4.3 Experimental Details
4.4 Results on MICCAI EndoVis 2017 Dataset
4.5 Results on Kvasir-Instrument
4.6 Ablation Studies
5 Conclusion and Discussion
References
Continual Domain Adaption for Neural Machine Translation
1 Introduction
2 Related Work
2.1 Continual Learning
2.2 Knowledge Distillation
3 Method
3.1 Multi-stage Incremental Framework
3.2 Analysis of KD
3.3 Alleviate the Negative Impact of KD
4 Experiments
4.1 Data Preparation
4.2 Baselines
4.3 Implementation Details
4.4 Main Results
4.5 Hyperparameters
4.6 Case Study
5 Conclusion
References
Neural-Symbolic Reasoning with External Knowledge for Machine Reading Comprehension
1 Introduction
2 Related Work
3 Methodology
3.1 Logic Extraction and Extension
3.2 Logical Graph Reasoning
3.3 Answer Prediction
4 Experiments
4.1 Dataset
4.2 Experimental Settings
4.3 Experimental Results
4.4 Ablation Study
4.5 Case Study
5 Conclusion
References
Partial Multi-label Learning via Constraint Clustering
1 Introduction
2 Related Work
3 Proposed Method
3.1 Pre-clustering
3.2 Feature Selection (FS)
3.3 Clustering
4 Experiment
4.1 Dataset
4.2 Metrics
4.3 Competitors
4.4 Experimental Results
4.5 Parameters Analysis
4.6 Time Complexity
5 Conclusion
.1 Proof of Formula
.2 Additional Excremental Result
References
Abstractive Multi-document Summarization with Cross-Documents Discourse Relations
1 Introduction
2 Related Work
2.1 Discourse Rhetorical Structure Based Multi-document Summarization
2.2 Abstractive Multi-document Summarization
3 Discourse Patterns Construction
3.1 Tree Generation Model
3.2 Discourse Patterns
4 Model Description
5 Experiments
5.1 Dataset and Evaluation Metrics
5.2 Training Details
5.3 Main Result
5.4 Ablation Study
6 Conclusion
References
MelMAE-VC: Extending Masked Autoencoders to Voice Conversion
1 Introduction
2 Related Work
3 MelMAE-VC
3.1 Pre-train Network MelMAE
3.2 Speaker Embedder
3.3 Style Transfer Decoder
4 Experiments
4.1 Conditions of Experiments
4.2 Objective Evaluation
4.3 Subjective Evaluation
5 Conclusions
References
Aspect-Based Sentiment Analysis Using Dual Probability Graph Convolutional Networks (DP-GCN) Integrating Multi-scale Information
1 Introduction
2 Related Work
3 Reshaped Syntactic Dependency Trees and Multi-scale Information
3.1 Aspect-Based Syntactic Dependency Tree Corresponding Syntactic Dependency Weights
3.2 Multi-scale Information
4 Proposed DP-GCN Model
4.1 Interactive Attention
4.2 Fusion of Multi-scale Information
4.3 Semantic Probabilistic Graph Convolution Module
4.4 Syntactic Probabilistic Graph Convolutional Module
4.5 Sentiment Classification
5 Experiments
5.1 Dataset and Evaluation Criteria
5.2 Parameter Setting
5.3 Baseline Methods
5.4 Experimental Results and Analysis
5.5 Ablation Experiment
6 Conclusion
References
Privacy-Preserving Image Classification and Retrieval Scheme over Encrypted Images
1 Related Work
2 The System Model and Security Assumption
2.1 System Model
2.2 Security Assumption
3 Image Classification and Privacy Protection
3.1 System Structure of PICR
3.2 Image Feature Extraction Model of PICR
3.3 Privacy Protection Algorithms
3.4 Privacy-Preserving Image Retrieval
4 Security Analysis of PICR
4.1 Security of Category Hash Code Encyption
4.2 Privacy Guarantee on Feature Vectors
5 Experimental Evaluation
5.1 Accuracy Comparison
5.2 Performance Evaluation and Comparison
6 Conclusion
References
An End-to-End Structure with Novel Position Mechanism and Improved EMD for Stock Forecasting
1 Introduction
2 Methodology
2.1 ACEFormer
2.2 ACEEMD
2.3 Time-Aware Mechanism
3 Experiments
3.1 Datasets
3.2 Model Setting
3.3 Evaluation Metrics
3.4 Competing Methods
4 Result
4.1 Trend Evaluation
4.2 ACEEMD Effect
5 Conclusion
References
Multiscale Network with Equivalent Large Kernel Attention for Crowd Counting
1 Introduction
2 Related Works
2.1 Multiscale Crowd Counting Methods
2.2 Attention-Based Methods
3 Proposed Approach
3.1 Large Kernel Attention Unit (LKAU)
3.2 Gate Channel Attention Unit (GCAU)
4 Implementation Details
4.1 Ground Truth Generation
4.2 Datasets
4.3 Training Details
5 Experiments
5.1 Evaluation Metrics
5.2 Comparisons with State-of-the-Art
5.3 Ablation Study
6 Conclusion
References
M3FGM: A Node Masking and Multi-granularity Message Passing-Based Federated Graph Model for Spatial-Temporal Data Prediction
1 Introduction
2 Related Work
2.1 Graph Neural Networks
2.2 Split Federated Learning
3 Problem Formulation
4 Methodology
4.1 Server Model
4.2 Client Model
4.3 Training and Inference Process
5 Experiments
5.1 Datasets
5.2 Compared Models and Settings
5.3 Performance Comparison Under the Ideal Scenario
5.4 Performance Comparison Under the Non-Ideal Scenario
5.5 Effect of Mask Node Rate and Discussion
6 Conclusion
References
LenANet: A Length-Controllable Attention Network for Source Code Summarization
1 Introduction
2 Method
2.1 Encoder with Length Offset Vector
2.2 Sentence-Level Code Tree
2.3 Context-Aware Code Sentence Extractor
3 Experiments
3.1 Experimental Setup
3.2 Baselines
3.3 Main Results
4 Discussion
4.1 Specific Feature Selection Without Length Offset
4.2 Visualization Analysis
5 Related Work
5.1 Code Summarization
5.2 Controllable Text Generation
6 Conclusions
References
Self-supervised Multimodal Representation Learning for Product Identification and Retrieval
1 Introduction
2 Related Work
2.1 Product Similarity
2.2 Multimodal Representation
3 Methodology
3.1 Self-attention Mechanism
3.2 Self-supervision Module
3.3 Multimodal Training Objectives
4 Experiments
4.1 Datasets
4.2 Evaluation Data and Metrics
4.3 Evaluation Results
4.4 Ablation Studies
5 Conclusion
References
Author Index