Computational Science – ICCS 2023: 23rd International Conference, Prague, Czech Republic, July 3–5, 2023, Proceedings, Part IV

Description

The five-volume set LNCS 14073-14077 constitutes the proceedings of the 23rd International Conference on Computational Science, ICCS 2023, held in Prague, Czech Republic, during July 3-5, 2023. The total of 188 full papers and 94 short papers presented in this book set were carefully reviewed and selected from 530 submissions. 54 full and 37 short papers were accepted to the main track; 134 full and 57 short papers were accepted to the workshops/thematic tracks. The theme for 2023, "Computation at the Cutting Edge of Science", highlights the role of Computational Science in assisting multidisciplinary research. This conference was a unique event focusing on recent developments in scalable scientific algorithms, advanced software tools; computational grids; advanced numerical methods; and novel application areas. These innovative novel models, algorithms, and tools drive new science through efficient application in physical systems, computational and systems biology, environmental systems, finance, and others. Preface Organization Contents – Part IV Computational Social Complexity The Social Graph Based on Real Data 1 Introduction and Model 2 Results and Conclusions References Longitudinal Analysis of the Topology of Criminal Networks Using a Simple Cost-Benefit Agent-Based Model 1 Introduction 2 Methodology 2.1 Agent-Based Modelling 2.2 Network Initialisation 2.3 Determine Topological Changes 2.4 Data 2.5 Measurement 3 Results and Discussion 4 Conclusion References Manifold Analysis for High-Dimensional Socio-Environmental Surveys 1 Introduction 2 Methods 2.1 Dimension Reduction Algorithms 2.2 Simulation Framework 2.3 Bangladesh Climate Change Adaptation Survey 3 Results 3.1 Simulation Framework Results 3.2 Bangladesh Climate Change Adaptation Survey Results 4 Discussion References Toxicity in Evolving Twitter Topics 1 Introduction 2 Related Work 3 Data and Methods 3.1 Topic Modelling and DAG Lineage 3.2 Transition Types 4 Studying Toxicity in Topic Evolution 4.1 Toxicity per Transition Type 4.2 Relationship Topic Popularity - Toxicity 5 Conclusion References Structural Validation of Synthetic Power Distribution Networks Using the Multiscale Flat Norm 1 Introduction 2 Methods 2.1 Multiscale Flat Norm 2.2 Proposed Algorithm 2.3 Normalized Flat Norm 3 Results and Discussion 3.1 Comparing Network Geometries 3.2 Comparison of Flat Norm and Hausdorff Distance Metrics 4 Conclusions References OptICS-EV: A Data-Driven Model for Optimal Installation of Charging Stations for Electric Vehicles 1 Introduction 1.1 Our Contributions 2 Related Work 3 Methodology 3.1 EV Charging Station Placement 3.2 Connecting EV Charging Stations 4 Experimental Results 4.1 EV Charging Station Placement 4.2 Optimal Routing Problem 5 Discussions and Conclusion References Computer Graphics, Image Processing and Artificial Intelligence Radial Basis Function Neural Network with a Centers Training Stage for Prediction Based on Dispersed Image Data 1 Introduction 2 Model and Methods 3 Datasets and Results 4 Conclusion References Database of Fragments of Medieval Codices of the 11th–12th Centuries – The Uniqueness of Requirements and Data 1 Introduction 2 Existing Databases 3 Dataset Description and Analysis 4 Conclusion and Future Research References Global Optimisation for Improved Volume Tracking of Time-Varying Meshes 1 Introduction 2 Related Work 3 As-Rigid-as-Possible Volume Tracking 4 Maximum Distance Based Affinity 5 Irregular Center Detection 6 Global Optimisation 6.1 Global Tracking Energy 6.2 Optimisation Strategy 6.3 Global Movement-Based Affinity 7 Experimental Results 7.1 Influence of the Proposed Affinity 7.2 Irregular Center Removal 8 Conclusions References Detection of Objects Dangerous for the Operation of Mining Machines 1 Introduction 2 Preparing a Dataset 3 Applied and Tested Models of Neural Networks 3.1 RetinaNet 3.2 Mask RCNN 3.3 YOLOv5 4 Detection System 5 Results 5.1 Evaluation of Networks 5.2 Evaluation of Created Detectors 6 Conclusion References A Novel DAAM-DCNNs Hybrid Approach to Facial Expression Recognition to Enhance Learning Experience 1 Introduction 2 Related Works 2.1 Complete Solutions for Face Expression Recognition 2.2 Applying Image Pre-processing Prior to CNNs Classification 2.3 Features Extraction Using CNNs Coupled with Other Machine Learning Classifiers 3 DAAM-DCNNs Hybrid Approach to Facial Expression Recognition 3.1 Convolutional Neural Network 4 Experimental Setup and Results 4.1 Datasets 4.2 Investigated Parameters 4.3 Results and Discussion 5 Conclusion References Champion Recommendation in League of Legends Using Machine Learning*-1pc 1 Introduction 1.1 Gameplay Overview 1.2 Pick and Ban Phase 2 Related Work 3 Methodology 3.1 Formulation of Machine Learning Problem 3.2 Datasets 3.3 Machine Learning Models for Solving the Problem 4 Results and Discussion 4.1 Pre-made Datasets 4.2 Riot API Datasets 4.3 Execution Time 5 Conclusions References Classification Performance of Extreme Learning Machine Radial Basis Function with K-means, K-medoids and Mean Shift Clustering Algorithms 1 Introduction 2 Extreme Learning Machine 3 Extreme Learning Machine Radial Basis Function 4 Clustering Methods 4.1 Mean Shift 4.2 K-means 4.3 K-medoids 5 Experiments and Results 6 Conclusions References Impact of Text Pre-processing on Classification Accuracy in Polish 1 Introduction 2 Related Works in Text Pre-processing on Classification Accuracy 3 Machine Translation Model for English-Polish Translation 3.1 Chosen Model Architecture 3.2 The Dataset Used for Machine Translation Task 3.3 Machine Translation Task Results 3.4 Summary of Machine Translation Task 4 Text Pre-processing Impact on Text 4.1 Polish Sentences Dataset Used in Classification 4.2 Development Tools Used for Performing Experiments 4.3 Experiments Verifying the Impact of Noise Removal 4.4 Results of the Noise Reduction Experiments 5 Conclusions References A Method of Social Context Enhanced User Preferences for Conversational Recommender Systems*-1pc 1 Introduction 2 Related Work 2.1 Recommender Systems "026E30F Conversational Recommender Systems 2.2 Social Context Information 3 Preliminaries 3.1 Problem Formulation 4 Methodology 4.1 Model Overview 4.2 Representation Learning 4.3 Social-Enhanced User Preference Estimation 4.4 Item and Attribute Scoring 4.5 Model Training 5 Experiments Setups 5.1 Datasets 5.2 Evaluation Metrics 5.3 Baselines 5.4 Implementation Details 6 Results and Discussion 6.1 Performance Comparison for Multi-round CRS 6.2 Performance Comparison at Different Conversation Turns 6.3 Ablation Study 6.4 Performance Comparison for User Preference Estimation 7 Conclusion References Forest Image Classification Based on Deep Learning and XGBoost Algorithm*-1pc 1 Introduction 2 Related Studies 3 Proposed Model 3.1 Multi-label Image Classification 3.2 Pre-processing 4 Overview of the Model Architecture 4.1 The XGBOOST Algorithm 4.2 ResNet50 Network Architecture 5 Metrics for the Study 6 Results and Discussion 7 Conclusion References Radius Estimation in Angiograms Using Multiscale Vesselness Function 1 Introduction 2 Methods 2.1 Vesselness-Radius Relationship 2.2 Curve Fitting to Estimate Vessel Radius 2.3 Reference Methods 3 Results 3.1 Radius Estimation Results in Images of Cylinders 3.2 Radius Estimation Results in Bifurcation Image 3.3 Radius Estimation in MRA 4 Summary and Conclusions References 3D Tracking of Multiple Drones Based on Particle Swarm Optimization 1 Introduction 2 A Method for Tracking Multiple Drones 2.1 Dataset 2.2 Particle Swarm Optimization 2.3 Fitness Function 3 Experiment Results 3.1 Simulation Dataset 3.2 Real Dataset 4 Conclusions References Sun Magnetograms Retrieval from Vast Collections Through Small Hash Codes 1 Introduction 2 Solar Magnetic Intensity Hash for Solar Image Retrieval 2.1 Magnetic Region Detection 2.2 Calculation of Solar Magnetic Intensity Descriptor 2.3 Hash Generation 2.4 Retrieval 3 Experimental Results 4 Conclusions References Cerebral Vessel Segmentation in CE-MR Images Using Deep Learning and Synthetic Training Datasets 1 Introduction 2 Related Work 2.1 State-of-the-Art Methods 2.2 Current Contribution 3 Methods and Materials 3.1 Vessel Segmentation Model 3.2 MR Angiography Simulation 4 Experimental Results 4.1 Simulated Training Images 4.2 Tests of the Segmentation Model 5 Conclusions References Numerical Method for 3D Quantification of Glenoid Bone Loss 1 Introduction 2 Statement of the Problem 3 Volume of a Polyhedron via the Gauss Formula 4 The Voxelization Approach 5 Description of Our Numerical Method 5.1 Randomized Sampling of the Surface 5.2 Downsampling the Point Cloud 5.3 Distance Function and Closest Point Map 5.4 Indicator Function 5.5 Denoising 5.6 Test of Accuracy 6 Numerical Illustration 7 Conclusion References Artificial Immune Systems Approach for Surface Reconstruction of Shapes with Large Smooth Bumps*-1pc 1 Introduction 1.1 Motivation 1.2 Aims and Structure of this Paper 2 Previous Work 3 The Optimization Problem 4 The Proposed Method: ClonalG Algorithm 5 Experimental Results 5.1 Example I 5.2 Example II 5.3 Example III 5.4 Implementation Issues 6 Conclusions and Future Work References Machine Learning and Data Assimilation for Dynamical Systems Clustering-Based Identification of Precursors of Extreme Events in Chaotic Systems 1 Introduction 2 Methodology 2.1 Preparatory Steps 2.2 Transition Probability Matrix and Graph Interpretation 2.3 Modularity-Based Clustering 2.4 Extreme and Precursor Clusters Identification 3 Results 3.1 MFE System 3.2 Kolmogorov Flow 4 Conclusions References Convolutional Autoencoder for the Spatiotemporal Latent Representation of Turbulence*-1pc 1 Introduction 2 Minimal Flow Unit 3 Multiscale Convolutional Autoencoder 4 Reconstruction Error 5 Conclusion References Graph Neural Network Potentials for Molecular Dynamics Simulations of Water Cluster Anions 1 Introduction 2 Background 3 Methods 4 Results and Discussion 5 Conclusions References Bayesian Optimization of the Layout of Wind Farms with a High-Fidelity Surrogate Model*-1pc 1 Introduction 2 Methodology 2.1 The Optimization Problem 2.2 Flow Solver 2.3 Bayesian Optimization Algorithm 3 Results 4 Conclusions References An Analysis of Universal Differential Equations for Data-Driven Discovery of Ordinary Differential Equations 1 Introduction 2 Related Work 3 Universal Differential Equations 4 UDE for Data-Driven Discovery of ODEs 5 Empirical Analysis 5.1 Training Procedure 5.2 Solver Accuracy 5.3 Functional Dependence and Data Sampling 6 Conclusions References Learning Neural Optimal Interpolation Models and Solvers 1 Introduction 2 Problem Statement and Related Work 3 Neural OI Framework 3.1 Neural OI Model and Solver 3.2 Learning Setting 3.3 Extension to Non-linear and Multimodal Optimal Interpolation 4 Experiments 4.1 2D+t GP Case-Study 4.2 Satellite Altimetry Dataset 5 Conclusion References Physics-Informed Long Short-Term Memory for Forecasting and Reconstruction of Chaos 1 Introduction 2 Chaotic Dynamical Systems 2.1 State Reconstruction 3 Physics-Informed Long Short-Term Memory 4 State Reconstruction and Lyapunov Exponents of the Lorenz-96 Model 5 Conclusions and Future Directions References Rules' Quality Generated by the Classification Method for Independent Data Sources Using Pawlak Conflict Analysis Model*-1pc 1 Introduction 2 Related Works 3 Model and Methods 4 Data Sets, Results and Discussion 4.1 Classification Quality 4.2 Rules' Quality 5 Conclusion References Data-Driven Stability Analysis of a Chaotic Time-Delayed System*-1pc 1 Introduction 2 Stability Analysis for Time-Delayed Systems 2.1 Time-Delayed Thermoacoustic System 3 Echo State Network 4 Results 5 Conclusion References Learning 4DVAR Inversion Directly from Observations 1 Introduction 2 Related Work 3 Data Assimilation and Learning Framework 3.1 State-Space System 3.2 The Initial Value Inverse Problem 3.3 Variational Assimilation with 4DVAR 3.4 Learning Inversion Directly from Observations 4 Experiments and Results 4.1 Lorenz96 Dynamics and Observations 4.2 Algorithm Benchmarks 4.3 Results 5 Conclusion References Human-Sensors & Physics Aware Machine Learning for Wildfire Detection and Nowcasting 1 Introduction 2 Background and Literature Review 3 Methods 4 Results: Wildfire Ignition and Spread Prediction 5 Conclusion and Future Work References An Efficient ViT-Based Spatial Interpolation Learner for Field Reconstruction 1 Introduction 2 Related Works and Contribution of the Present Work 3 Methodology 4 Test Cases and Results 5 Conclusion References A Kernel Extension of the Ensemble Transform Kalman Filter 1 Introduction 2 ETKF Reformulation with Kernel Methods 2.1 ETKF Formulation and Classical Resolution 2.2 Reformulation and Resolution of ETKF with Kernels 3 Numerical Experiments 3.1 Experimental Setup 3.2 Discussion 4 Conclusion and Perspectives A Construction of Pa When K is Invertible References Fixed-Budget Online Adaptive Learning for Physics-Informed Neural Networks. Towards Parameterized Problem Inference 1 Introduction 2 Methodology 2.1 Physics-Informed Neural Networks 2.2 Adaptive Learning Strategy for PDEs Residuals 3 Numerical Results 3.1 Burgers Equation 3.2 Application to Calendering Process 4 Conclusion References Towards Online Anomaly Detection in Steel Manufacturing Process 1 Introduction 2 Related Works 2.1 Data Streams 2.2 Anomaly Detection 2.3 Cold Rolling Process 3 Research Methods 3.1 Dataset Description 3.2 Dealing with Imbalanced Data 3.3 Learning Scenarios 3.4 Learning Algorithms and Validation 3.5 Anomaly Detection 4 Results and Discussion 5 Conclusion and Future Works References MeshFree Methods and Radial Basis Functions in Computational Sciences Biharmonic Scattered Data Interpolation Based on the Method of Fundamental Solutions 1 Introduction 2 Biharmonic Interpolation 3 The Method of Fundamental Solutions Applied to the Biharmonic Interpolation Problem 3.1 Solution of the Biharmonic Problem by Overlapping Schwarz Method 3.2 Localization of the MFS for the Biharmonic Equation Based on Overlapping Schwarz Method 3.3 Localized Solution of the Biharmonic Interpolation Problem 4 A Numerical Example 5 Conclusions References Spatially-Varying Meshless Approximation Method for Enhanced Computational Efficiency*-1pc 1 Introduction 2 Numerical Treatment of Partial Differential Equations 2.1 Computational Stability 2.2 Implementation Details 3 Governing Problem 4 Numerical Results 4.1 The de Vahl Davis Problem 4.2 Natural Convection on Irregularly Shaped Domains 4.3 Application to Three-Dimensional Irregular Domains 5 Conclusions References Oscillatory Behaviour of the RBF-FD Approximation Accuracy Under Increasing Stencil Size 1 Introduction 2 Problem Setup 3 Results 4 Conclusions References On the Weak Formulations of the Multipoint Meshless FDM 1 Introduction 2 Multipoint Problem Formulation 3 Weak Formulations of the Multipoint MFDM Approach 4 Numerical Analysis 4.1 Benchmark Tests 4.2 Weak and Strong Formulations in Nonlinear and Multiscale Analyses 5 Final Remarks References Multiscale Modelling and Simulation Convolutional Recurrent Autoencoder for Molecular-Continuum Coupling*-1pc 1 Introduction 2 Molecular-Continuum Coupled Flow 2.1 Dataset Creation 3 Convolutional Recurrent Autoencoder 3.1 Convolutional Autoencoder (AE) 3.2 Recurrent Neural Network 3.3 Hybrid Model 4 Implementation and Training Approach 5 Results – Couette Flow Scenario 6 Results – Kármán Vortex Street Scenario 7 Conclusions References Developing an Agent-Based Simulation Model to Forecast Flood-Induced Evacuation and Internally Displaced Persons 1 Introduction 2 Literature Review 2.1 Evacuation Modelling Approaches 3 Development Approach 3.1 Assumptions 3.2 Conceptual Model 3.3 Model Inputs 4 A Case Study: Nigeria, Bauchi State 5 Results and Discussion 6 Conclusion References Towards a Simplified Solution of COVID Spread in Buildings for Use in Coupled Models 1 Introduction 2 Conceptual Model 3 Establishing a Realistic Base Infection Probability 4 Multiscale Simulation Approach 5 Showcase 6 Discussion References Epistemic and Aleatoric Uncertainty Quantification and Surrogate Modelling in High-Performance Multiscale Plasma Physics Simulations 1 Introduction 2 Methodology 2.1 Epistemic Uncertainty 2.2 Aleatoric Uncertainty 2.3 Surrogate Modelling 3 Numerical Results 3.1 Computational Model 3.2 Simulations and Resulting Time Traces 3.3 Aleatoric Uncertainty 3.4 Epistemic Uncertainty 3.5 Surrogate Model 4 Discussion References Network Models and Analysis: From Foundations to Complex Systems Applying Reinforcement Learning to Ramsey Problems*-1pc 1 Introduction 2 Notation and Definitions Related to Graphs and Ramsey Numbers 3 RLS Applied to Determining Critical Colorings for Some Ramsey Numbers 4 Computational Experiments 5 Conclusion References Strengthening Structural Baselines for Graph Classification Using Local Topological Profile*-1pc 1 Introduction 2 Related Works 3 Methods 4 Results and Discussion 5 Conclusions References Heuristic Modularity Maximization Algorithms for Community Detection Rarely Return an Optimal Partition or Anything Similar 1 Introduction 2 Methods and Materials 2.1 Modularity 2.2 Modularity Maximization 2.3 Sparse IP Formulation of Modularity Maximization 2.4 Reviewing Eight Heuristic Modularity Maximization Algorithms 2.5 Measures for Evaluating Heuristic Algorithms 2.6 Data and Resources 3 Results 3.1 Comparing Partitions from Different Algorithms on One Network 3.2 Multiplicity of Optimal Partitions 3.3 Evaluating Heuristic Algorithms on 80 Networks 3.4 Success Rate of Heuristic Algorithms in Maximizing Modularity 4 Discussions and Future Directions References Analyzing the Attitudes of Consumers to Electric Vehicles Using Bayesian Networks 1 Introduction 2 Data Collection, the Sample and Methods 3 Analysis Using a Bayesian Network 4 Results and Discussion 5 Limitations of the Study 6 Conclusions and Future Work References Parallel Triangles and Squares Count for Multigraphs Using Vertex Covers*-1pc 1 Introduction 2 Notation 3 Computation of Vertex Covers 4 Counting Triangles 4.1 Per Node Triangle Count 5 Counting Squares 5.1 Per Node Version 6 Experiments 6.1 Impact of Vertex Cover Schema 6.2 Scalability 7 Future Works 8 Conclusions References Deep Learning Attention Model for Supervised and Unsupervised Network Community Detection 1 Introduction 2 Methodology 3 Results 4 Conclusions References On Filtering the Noise in Consensual Communities 1 Introduction 2 Definitions 3 Related Work 3.1 The Consensus Matrix 3.2 From Consensus Matrix to Consensual Communities 3.3 Complexity Issues 4 Information is Noisy 4.1 Graph Distance 4.2 Edge Clustering Coefficient 5 Finding an Optimum Number of Entries – Filtering Out The noise 5.1 Improving Existing Algorithms 6 Experiments 6.1 Synthetic Graphs 6.2 Limitations 6.3 Real Graphs 7 Conclusion References Author Index