Computational Science – ICCS 2022: 22nd International Conference, London, UK, June 21–23, 2022, Proceedings, Part III

Description

The four-volume set LNCS 13350, 13351, 13352, and 13353 constitutes the proceedings of the 22ndt International Conference on Computational Science, ICCS 2022, held in London, UK, in June 2022.* The total of 175 full papers and 78 short papers presented in this book set were carefully reviewed and selected from 474 submissions. 169 full and 36 short papers were accepted to the main track; 120 full and 42 short papers were accepted to the workshops/ thematic tracks. *The conference was held in a hybrid format Preface Organization Contents – Part III Knowledge Discovery in Databases: Comorbidities in Tuberculosis Cases-5pt 1 Background 1.1 The TB Scenario in Brazil and in the World 1.2 The Comorbidities in TB Cases 1.3 Objectives 2 Methods 3 Results and Discussion 3.1 Data Characterization 3.2 Pre-processing 3.3 Analysis: FS Filter Method + Rules Extraction 3.4 Analysis: FS Wrapper Approaches + Classification Algorithms 4 Conclusion References Noninvasive Estimation of Mean Pulmonary Artery Pressure Using MRI, Computer Models, and Machine Learning 1 Introduction 2 Related Work 3 Methods 3.1 PH Dataset 3.2 Features Derived from Models of Blood Circulation 3.3 Machine Learning for PH Detection 4 Results 4.1 mPAP Regression 4.2 PH Classification 5 Discussion 6 Conclusion 7 Appendix References GAN-Based Data Augmentation for Prediction Improvement Using Gene Expression Data in Cancer 1 Introduction 2 Methodology 2.1 Noise Injection Method 2.2 SMOTE Techniques 2.3 Conditional Generative Adversarial Networks 2.4 CGAN Modified Generative Process 2.5 Modifications for Treating Unbalanced Data Set Distribution 2.6 Implementations of the Models 2.7 Benchmark Data Sets 3 Experiments and Results 4 Conclusion References National Network for Rare Diseases in Brazil: The Computational Infrastructure and Preliminary Results 1 Introduction 2 Literature Review 3 Methods 3.1 Study Design 3.2 Electronic Data Captures Systems and Data Collection 3.3 Web Portal Development and Data Analysis 4 Results and Discussion 5 Conclusion References Classification of Uterine Fibroids in Ultrasound Images Using Deep Learning Model 1 Introduction 2 Proposed Classification Method 3 Result and Discussion 4 Conclusion References Explainable AI with Domain Adapted FastCAM for Endoscopy Images 1 Introduction 2 Related Work 3 Domain Adapted (DA)-FastCAM 4 Optimization Approach for DA-FastCAM 4.1 Optimization Framework and Algorithms 5 Evaluation of DA-FastCAM 5.1 Optimization of Weights for Grasper and Hook Together 5.2 Optimization of Weights for Grasper 5.3 Optimization of Weights for Hook 5.4 Optimization of Weights for Grasper of Frame 312 5.5 Optimization of Weights for Hook of Frame 28926 5.6 Optimization Result Overview 6 Conclusion References Sensitivity Analysis of a Model of Lower Limb Haemodynamics 1 Introduction 2 Methods 2.1 Lower Limb Circulation Model 2.2 Sensitivity Analysis 3 Results 4 Discussion 5 Conclusion References Data Augmentation Techniques to Improve Metabolomic Analysis in Niemann-Pick Type C Disease 1 Introduction 2 Materials 3 Data Augmentation Methods 4 Experiments and Results 4.1 Classification Performance 4.2 Augmented Datasets Analysis 5 Conclusions References Effect of Feature Discretization on Classification Performance of Explainable Scoring-Based Machine Learning Model 1 Introduction 2 Related Research 3 Methods 3.1 Risk-calibrated Supersparse Linear Integer Model (RiskSLIM) 3.2 RiskSLIM with Feature Discretization (FD-RiskSLIM) 3.3 Implementation 4 Experiments 4.1 Prediction of Death of Patients with Heart Failure 4.2 Prediction of Outcomes of Kidney Cancer Treatment 5 Conclusions 7 Appendix References Hybrid Modeling for Predicting Inpatient Treatment Outcome: COVID-19 Case 1 Introduction 2 Model 2.1 Data Mining 3 Experimental Study 3.1 Mortality Prediction – Bayesian Networks 3.2 Prediction Experiment with Critically Ill Patients 3.3 Predictive Modeling with Machine Learning Algorithms 4 Conclusion and Future Work References Neural Additive Models for Explainable Heart Attack Prediction 1 Introduction 2 Related Works 3 Methods 4 Experiments 5 Results and Discussion 6 Conclusion References Machine Learning Models for Predicting 30-Day Readmission of Elderly Patients Using Custom Target Encoding Approach 1 Introduction 2 Study Design and Methodology 2.1 Data Preparation 2.2 Data Pre-processing 2.3 Modelling 2.4 Model Evaluation 3 Results 4 Discussion References Patient- and Ventilator-Specific Modeling to Drive the Use and Development of 3D Printed Devices for Rapid Ventilator Splitting During the COVID-19 Pandemic 1 Introduction 2 Methods 2.1 Design of the Ventilator Splitter and Resistor System 2.2 Reduced Order Ventilator Computational Model 2.3 Benchtop Testing of VSRS Using Test Lungs 3 Results 3.1 The Computational Model Predicts Benchtop Measurements 3.2 Pressure-controlled Ventilation Protects Patients from Changes in the Opposing Patients Circuit 3.3 Tidal Volumes from Pressure-Controlled Ventilation Are Highly Sensitive to Small Changes in Ventilator Settings, Endotracheal Tube Sizes, and Patient Characteristics 3.4 The Computational Model Solves the Large Parameter Space of Potential Patient Pairings 4 Discussion and Conclusion References AI Classifications Applied to Neuropsychological Trials in Normal Individuals that Predict Progression to Cognitive Decline 1 Introduction 2 Methods 2.1 Rough Set Theory 3 Results 3.1 Statistics 3.2 Rules from General Model (G Model) 4 Discussion 5 Conclusions References Super-Resolution Convolutional Network for Image Quality Enhancement in Remote Photoplethysmography Based Heart Rate Estimation 1 Introduction 2 Related Works 3 Methods 3.1 Super Resolution Preprocessing 4 Results and Discussions 4.1 HR Estimation 5 Conclusion References A Hybrid Modeling Framework for City-Scale Dynamics of Multi-strain Influenza Epidemics 1 Introduction 2 Methods 2.1 The Multi-agent Model 2.2 The Compartmental Model 2.3 Switching Algorithm 3 Simulation 3.1 Switching Moment Influence 3.2 Protected Individuals Fraction Influence 4 Conclusions References Numerical Stability of Tangents and Adjoints of Implicit Functions 1 Introduction 2 Prerequisites 3 Errors in Tangents and Adjoints of Implicit Functions 3.1 Systems of Linear Equations 3.2 Systems of Nonlinear Equations 3.3 Convex Unconstrained Objectives 4 Conclusion References Analysis of Parameters Distribution of EEG Signals for Five Epileptic Seizure Phases Modeled by Duffing Van Der Pol Oscillator 1 Introduction 2 Materials and Methods 2.1 EEG Signals 2.2 Duffing Van Der Pol Oscillator 2.3 Optimization Scheme 3 Results 4 Discussion and Conclusion References Multi-criterial Design of Antennas with Tolerance Analysis Using Response-Feature Predictors 1 Introduction 2 Multi-criterial Antenna Optimization with Tolerance Analysis 2.1 Problem Formulation 2.2 Yield Estimation by Means of Response Features 2.3 Generating Trade-off Designs 2.4 Optimization Algorithm 3 Verification Case Studies 3.1 Case I: Dual-Band Dipole Antenna 3.2 Case II: Quasi-yagi Antenna 4 Conclusions References Expedited Optimization of Passive Microwave Devices Using Gradient Search and Principal Directions 1 Introduction 2 Microwave Design Optimization Using EM Models 2.1 Optimization Task Formulation 2.2 Gradient-Based Search with Numerical Derivatives 3 Accelerated Optimization by Means of Principal Directions 3.1 Principal Directions and Their Identification 3.2 Restricted Sensitivity Updates 3.3 Optimization Algorithm 4 Demonstration Case Studies and Benchmarking 4.1 Verification Circuits 4.2 Experimental Setup 4.3 Results and Discussion 5 Conclusions References Approach to Imputation Multivariate Missing Data of Urban Buildings by Chained Equations Based on Geospatial Information 1 Introduction 2 Techniques for Handling the Missing Data 2.1 Missing Data Types 2.2 Overview of Techniques for Handling the Missing Data 2.3 Machine Learning Methods Used in Data Imputation 3 Approach to Imputation Multivariate Missing Urban Buildings Data Based on Geospatial Information 3.1 Description of the Developed Approach 3.2 Findings of the Developed Approach 4 Imputation Missing Values to Saint-Petersburg Buildings Dataset 4.1 Collection and Analysis of Initial Dataset 4.2 Investigating the Effectiveness of an Approach to City Building Data Imputation 4.3 Comparative Analysis of Approaches to Data Imputation 5 Conclusion References Global Design Optimization of Microwave Circuits Using Response Feature Inverse Surrogates 1 Introduction 2 Global Design by Response-Feature Inverse Surrogates 2.1 Simulation-Based Optimization. Problem Formulation 2.2 Response Features for Design Assessment 2.3 Global Search Stage Using Inverse Models 2.4 Local Tuning 2.5 Algorithm Summary 3 Verification Experiments 3.1 Verification Circuits 3.2 Results 4 Conclusions References Classification of Soil Bacteria Based on Machine Learning and Image Processing 1 Introduction 2 Work-flow Scheme 2.1 Segmentation of the Region of Interest 2.2 Feature Generation 2.3 Feature Selection 2.4 Class Recognition 3 Features Based on Shape 3.1 Planar Cubic Spline Interpolation 3.2 Curvature Calculation 3.3 Features Calculation 4 Features Based on Texture 4.1 GLCM Features 4.2 GLRLM Features 5 Selected Class Recognition Methods 5.1 KNN 5.2 ELM-RBF 6 Experiments and Results 7 Conclusions References Calibration Window Selection Based on Change-Point Detection for Forecasting Electricity Prices 1 Introduction 2 The Data 3 Calibration Window Selection Using NOT 4 Forecasting Models 5 Results 6 Conclusions and Discussion References Tackling Air Pollution in Cities with Modelling and Simulation: Remote Group Model Building as an Educational Tool Supporting System Dynamics Modelling 1 Introduction 2 Related Works 3 Methodology 4 Elicitation of the General Causal Structures 5 Outline of the Initial Causal Structures with Discussion 6 Conclusions and Future Works References Fast Isogeometric Analysis Simulations of a Process of Air Pollution Removal by Artificially Generated Shock Waves 1 Introduction 2 Experimental Verification 3 Numerical Simulations 3.1 Fast Simulation of Thermal Inversion and Cloud Formation 3.2 Fast Simulation of the Pollution Removal by Artificially Generated Shock Waves 4 Comparison of Higher-Order and Continuity Isogeometric Finite Element Method with Finite Difference Method 5 Conclusion References Automatic Generation of Individual Fuzzy Cognitive Maps from Longitudinal Data 1 Introduction 2 Background 2.1 Fuzzy Cognitive Maps 2.2 Genetic Algorithms for Fuzzy Cognitive Maps 3 Methods 3.1 Proposed Algorithm to Generate an Individual FCM 3.2 Algorithms to Generate a Population from Longitudinal Data 4 Case Study 4.1 Overview 4.2 Comparing Algorithm 2 with the Original RCGA Procedure 4.3 One-for-Each vs. One-Size-Fits-All 5 Discussion References A Simulation Study of the Delayed Effect of Covid-19 Pandemic on Pensions and Welfare of the Elderly: Evidence from Poland 1 Introduction 2 The Polish Pension System 3 Simulation Methods in Retirement Planning 4 Methodology 4.1 Method and Output Measures 4.2 Assumptions 4.3 Input Distributions 4.4 Simulation Experiments 5 Simulation Results 5.1 Gross Replacement Rate (GRR) 5.2 Probability of Target Replacement Rates (TRR) 5.3 Delayed Retirement 5.4 The Comparative Analysis of Simulation Techniques 6 Conclusions References Non-Generic Case of Leap-Frog for Optimal Knots Selection in Fitting Reduced Data 1 Introduction 2 Preliminaries 3 Non-generic Case: First Acceleration and Last Velocity 4 Special Conditions for Non-generic Case of Leap-Frog 5 Perturbed Special Case 6 Conclusions References Adaptive Surrogate-Assisted Optimal Sailboat Path Search Using Onboard Computers 1 Introduction 2 Related Research 3 Problem Formulation 4 Proposed Solution 4.1 The Solution Space Representation 4.2 Surrogate-Based Performance Measure 4.3 Surrogate-Assisted Optimal Path Search Algorithm 5 Results and Discussion 6 Conclusions References Local Search in Selected Crossover Operators 1 Introduction 2 Crossover Operators 3 The Local-Search Enhancements of the Crossover Operators 3.1 Local Search Implementation in Conflict and Non-conflict Resolvers 4 Experimental Evaluation 4.1 The Example of Warehouse Optimization Problem 4.2 The Example of Traveling Salesman Problem 4.3 Experimental Setup and Parameters 4.4 Analysis of the Results Obtained for the Warehouse Problem 4.5 Analysis of the Results Obtained for TSP 5 Conclusions References Numerical and Statistical Probability Distribution Transformation for Modeling Traffic in Optical Networks 1 Introduction 2 Problem Formulation and Methods 2.1 Demands Matrix Elements Probability Distributions 2.2 Calculation of Edge Data Transmission Rate Probability Distributions for a Small Network 2.3 Extending Edge Calculations to Complex Networks 3 Results 4 Conclusions References GPU Accelerated Modelling and Forecasting for Large Time Series 1 Introduction 2 Recursive Schur Complement Based Time - Series Modelling 2.1 Assessment and Selection of Basis Functions 3 Performance Optimization and Parallelization 3.1 Graphics Processing Unit Acceleration 4 Numerical Results 4.1 Time Series 1 - Scalability and Accuracy 4.2 Time Series 2 - Flexibility and Higher Order Interactions 5 Conclusion References Intersection Representation of Big Data Networks and Triangle Enumeration 1 Introduction 1.1 k-count Distribution 2 Intersection Representation of Network Data 2.1 Adjacency Matrix-Based Triangle Counting 2.2 Intersection Matrix-Based Triangle Counting 2.3 Data Structure 2.4 Local Triangle Count and Edge Support 2.5 Algorithm 3 Numerical Results 4 Conclusion References Global Surrogate Modeling by Neural Network-Based Model Uncertainty 1 Introduction 2 Methods 3 Numerical Experiments 3.1 One-Dimensional Forrester Function 3.2 Two-Dimensional Branin Function 4 Conclusion References Analysis of Agricultural and Engineering Systems Using Simulation Decomposition 1 Introduction 2 Methods 2.1 Problem Statement 2.2 Simulation Decomposition 3 Numerical Examples 3.1 Model Problem 3.2 Ultrasonic Testing (UT) System with a Pillbox Void 3.3 Iowa Food-Energy-Water System 4 Conclusion References Neural Network-Based Sequential Global Sensitivity Analysis Algorithm 1 Introduction 2 Methods 3 Numerical Experiments 3.1 Case 1: Analytical Function 3.2 Case 2: Ultrasonic Testing (UT) of a Pillbox-Defect 4 Conclusion References Development of an Event-Driven System Architecture for Smart Manufacturing 1 Introduction 2 Overview of the Tweeting Factory 2.1 Data Acquisition 2.2 From Tweets to Decisions 2.3 Tweeting Factory Architecture 3 Experiments 3.1 KPI Indexes Calculation by the Legacy Method 3.2 A New Data Acquisition Method 3.3 Simulation Implementation 3.4 Simulation Run 4 Conclusions References Boundary Geometry Fitting with Bézier Curves in PIES Based on Automatic Differentiation 1 Introduction 2 Defining the Boundary by Bézier Curves in PIES 3 Proposed Algorithm 4 Results and Evaluation 5 Conclusions References Partitioning Dense Graphs with Hardware Accelerators 1 Introduction 2 Graph Partitioning Formulations 3 Computational Experiments References A Taxonomy Guided Method to Identify Metaheuristic Components 1 Introduction 2 Literature Review 3 Taxonomy Guided Identification of Metaheuristic Components: TAXONOG-IMC 3.1 Comprehensive Description of Method Process 4 Application of Method 4.1 Behavior Taxonomy 4.2 Natural-Inspiration Taxonomy 4.3 Dimension Descriptions 5 Analysis and Discussion 6 Conclusion References Camp Location Selection in Humanitarian Logistics: A Multiobjective Simulation Optimization Approach 1 Introduction 1.1 Related Work 2 A Multiobjective Camp Location Selection Model 2.1 A Simulation-Optimization Approach 3 Test Setup and Results 4 Conclusion References A Sparse Matrix Approach for Covering Large Complex Networks by Cliques 1 Introduction 2 Compact Representation and Edge Clique Cover 2.1 Intersection Representation 3 An Edge-Centric MinECC Algorithm 3.1 Selection of Uncovered Edges 3.2 The Algorithm 3.3 Discussion 4 Numerical Testing 5 Conclusion References DSCAN for Geo-social Team Formation 1 Introduction 2 Related Work 3 Problem Definition 4 Our Framework 4.1 Network Distribution 4.2 Suitable Cluster Selection 4.3 Geo-social Team Formation 4.4 Complexity Analysis 5 Experimental Analysis 5.1 Experiment Setup 5.2 Effectiveness Evaluation 5.3 Efficiency Evaluation 6 Conclusions References Data Allocation with Neural Similarity Estimation for Data-Intensive Computing 1 Introduction 2 State of the Art 3 Methods 3.1 Background 3.2 Neural Similarity Estimator 3.3 Data Allocation Model 4 Justification and Evaluation 5 Conclusion and Outlook References Your Social Circle Affects Your Interests: Social Influence Enhanced Session-Based Recommendation 1 Introduction 2 Related Works 3 Modeling Methods 3.1 Problem Definition 3.2 Social Influence Modeling 3.3 Model Architecture and Training 4 Experiments 4.1 Experimental Setup 4.2 Experimental Evaluations 5 Conclusion References Action Recognition in Australian Rules Football Through Deep Learning 1 Introduction 2 Literature Review 3 Australian Rules Football Data Set 4 Implementation and Discussion of Results 5 Team Identification and Associated Limitations 6 Conclusions and Future Work References SEGP: Stance-Emotion Joint Data Augmentation with Gradual Prompt-Tuning for Stance Detection 1 Introduction 2 Related Work 2.1 Stance Detection 2.2 Data Augmentation 2.3 Curriculum Learning 2.4 Prompt-Tuning 3 Method 3.1 Preliminaries 3.2 Overall Architecture 3.3 Stance-Emotion Joint Data Augmentation 3.4 Gradual Prompt-Tuning 4 Experiment 4.1 Dataset and Baseline Methods 4.2 Experimental Results 4.3 Analysis of Stance-Emotion Joint Data Augmentation 4.4 Analysis of Gradual Prompt-tuning 5 Conclusion References Image Features Correlation with the Impression Curve for Automatic Evaluation of the Computer Game Level Design 1 Introduction 2 Image Features 3 Evaluation 4 Results and Analysis 4.1 Individual Image Features Correlation 4.2 Combined Image Features Correlation 4.3 Best Features for Different Level Design Stages 5 Conclusion References ACCirO: A System for Analyzing and Digitizing Images of Charts with Circular Objects 1 Introduction 2 The Workflow of ACCirO 3 Experiments and Results 4 Conclusions References Learning Scale-Invariant Object Representations with a Single-Shot Convolutional Generative Model 1 Introduction 2 Related Works 3 Method 3.1 The Proposed Model: SSDIR 4 Experiments 4.1 Per-object Reconstructions 4.2 Latent Space 4.3 Ablation Study 5 Conclusions References How to Sort Them? A Network for LEGO Bricks Classification 1 Introduction 2 Network Topologies 3 The Dataset 4 Training Process 5 The Results 5.1 Stage I - The Four-Hour Training 5.2 Stage II - The Twelve-Hour Training 5.3 Final Tests 6 Conclusions References Novel Photoplethysmographic Signal Analysis via Wavelet Scattering Transform 1 Introduction 2 Materials and Methods 2.1 The Wavelet Scattering Transform 2.2 Classification Model 2.3 PPG Dataset 3 Results 4 Conclusions References Convolutional Neural Network Compression via Tensor-Train Decomposition on Permuted Weight Tensor with Automatic Rank Determination 1 Introduction 2 Preliminary 3 Proposed Method 3.1 Model 3.2 TT Decomposition Algorithm 3.3 Implementation 3.4 Computational Complexity 4 Results 4.1 CIFAR-10 4.2 CIFAR-100 5 Conclusions References Comparing Explanations from Glass-Box and Black-Box Machine-Learning Models 1 Introduction 2 Related Works and Motivation 3 Experimental Comparison 4 Results and Discussion 5 Summary References Virtual Reality Prototype of a Linear Accelerator Simulator for Oncological Radiotherapy Training 1 Introduction 2 Previous Work 2.1 Virtual Simulation for Medical Science 2.2 Oncology Radiotherapy Training Issues 2.3 VR for Radiotherapy Education and Training 3 System Design and Implementation 3.1 On-site Medical Facility Visits and Meetings with Experts 3.2 Functional Requirements 3.3 System Design 3.4 Implementation 4 Results 4.1 Remote Control Room 4.2 ORTLINAC Radiotherapy Room 4.3 User Interface and Interaction 5 Conclusions and Future Work References A Review of 3D Point Clouds Parameterization Methods 1 Introduction 2 Some Concepts 3 Parameterization Methods of Organized Point Clouds 4 Parameterization Methods of Unorganized Point Clouds 4.1 Base Surfaces-Based Methods 4.2 Meshless Parameterization 4.3 Spherical Mapping 4.4 Methods Adapted from Mesh Parameterization 4.5 Neural Networks-Based Methods 4.6 Other Methods 5 Conclusion References Statistical Prediction of Extreme Events from Small Datasets 1 Introduction 2 A Low-Dimensional Model for Turbulent Shear Flow 3 Echo State Networks 4 Statistical Prediction of Extreme Events 5 Conclusions and Future Directions References Outlier Detection for Categorial Data Using Clustering Algorithms 1 Introduction 2 State of Art 3 Data Clustering 3.1 K-modes Clustering 3.2 ROCK Clustering 4 Conducted Research 4.1 Data Description 4.2 Methodology 5 Experiments 5.1 Time Complexities of Clustering Algorithms 5.2 Outlier Detection for Clustering Results 5.3 Detection of Common Outliers 5.4 Evaluation of the Proposed Methods 6 Conclusions References Augmenting Graph Inductive Learning Model with Topographical Features 1 Introduction 2 Related Work 3 Problem Definition and Proposed Approach: 3.1 Subgraph Extraction 3.2 Subgraph Node Labeling 3.3 Feature Matrix Construction 3.4 Scoring Subgraph Using GNN 4 Experiments 4.1 Inductive Relation Prediction 4.2 Transductive Relation Prediction 5 Conclusions References Generative Networks Applied to Model Fluid Flows 1 Introduction 2 Methodology 2.1 POD-based Non-Intrusive Reduced Order Modelling 2.2 Generative Models 2.3 Space-time Predictions Using Generative Models 3 Implementation and Results 3.1 Case Study: Parameter-varying Flow in a Periodic Box 3.2 POD Compression and Order Reduction 3.3 Prediction Using GAN and AAE 4 Discussion and Conclusion References Towards Social Machine Learning for Natural Disasters 1 Introduction 2 Background 3 System Overview 3.1 [A] System Input/Setup 3.2 [B] Social Media Data Streaming 3.3 [C] Filtering and Text Pre-processing 3.4 [D] Information Extraction and Sentimental Analysis 3.5 [E] Data Persistence 3.6 [F] System Training (Pre-Operational) 3.7 [G] ML Prediction Model 4 Experimental Test Case: Wildfires - Satellites and Twitter 5 Conclusion and Future Work References Author Index