Artificial intelligence and soft computing : 20th international conference, ICAISC 2021, virtual event, June 21-23, 2021 : proceedings

Description

Preface Organization Contents – Part II Contents – Part I Computer Vision, Image and Speech Analysis Classification of Dermatological Asymmetry of the Skin Lesions Using Pretrained Convolutional Neural Networks 1 Introduction 1.1 Dermatological Asymmetry of Skin Lesions and Screening Methods 1.2 Dermatological Datasets 1.3 Pretrained Convolutional Neural Network and Their Features 2 Data Preparation for the Research 2.1 Augmentation and Preparation of the Database 2.2 CNN Network Setting and Configuration 2.3 Hardware Description 3 Research Method Description 4 Results 5 Conclusions References Contextual Image Classification Through Fine-Tuned Graph Neural Networks 1 Introduction 2 Background 3 Proposed Approach 4 Experiments 4.1 Scenarios 4.2 Dataset Description 4.3 Results 5 Conclusions References Architecture Monitoring and Reliability Estimation Based on DIP Technology 1 Introduction 2 Automatic Detection Algorithm 2.1 Grayscale Image Closing 2.2 Image Enhancement 2.3 Crack Regions Rapid Extraction 2.4 Morphological Noise Removal 2.5 Crack Areas Calculation 2.6 Crack Perimeter Calculation 2.7 Crack Circularity Calculation 3 Preliminary Experimental Result Comparison 4 Conclusions References An Efficient Technique for Filtering of 3D Cluttered Surfaces 1 Introduction 2 Cluttered Surface Filtering 2.1 Plane Surface Filtering 2.2 Euclidean Clustering 2.3 Computation of Features 2.4 Classification of Cluttered and Non-cluttered Surfaces 3 Experimental Evaluation 4 Conclusion References A Computer Vision Based Approach for Driver Distraction Recognition Using Deep Learning and Genetic Algorithm Based Ensemble 1 Introduction 2 Related Works 3 Dataset Information 3.1 American University of Cairo (AUC) Distracted Driver (V1) Dataset 3.2 State Farm Driver Distraction Dataset 4 Proposed Methodology 4.1 Independent Classifier Branches 4.2 Genetic Algorithm(GA) Based Ensemble 5 Experiments and Results 5.1 Results on AUC Driver Distraction (V1) Dataset 5.2 Results on State Farm Driver Distraction Dataset 6 Conclusion References Multimodal Image Fusion Method Based on Multiscale Image Matting 1 Introduction 2 The Proposed Multiscaling Image Matting (MSIM) Technique 2.1 Preprocessing 2.2 Feature Extraction 2.3 Multiscaling 3 Objective Evaluation Metrics 3.1 Mutual Information (MI) 3.2 Entropy (EN) 3.3 Feature Mutual Information (FMI) 3.4 Spatial Structural Similarity (SSS) QAB/F 3.5 Visual Information Fidelity (VIF) 4 Results and Discussion 4.1 Experimental Setup 4.2 Fusion Results 5 Conclusions References Targeting the Most Important Words Across the Entire Corpus in NLP Adversarial Attacks 1 Introduction 2 Related Works 3 Technical Description 3.1 Benchmark Datasets 3.2 Choosing and Targeting Words to Attack 3.3 Word Perturbation 3.4 Comparison of Methods and Evaluation 4 Experimental Results 4.1 Evaluation on IMDb 4.2 Evaluation on Stack Overflow 4.3 Transfer Learning from IMDb to Stack Overflow 4.4 Run Time Comparison 4.5 Results Analysis 5 Conclusion and Future Work References RGB-D Odometry for Autonomous Lawn Mowing 1 Introduction 2 Related Work 2.1 Algorithms Overview 3 Problem Description 3.1 Dataset Overview 3.2 Error Metrics 4 Research and Results 4.1 Errors Handling 5 Conclusions References Using PMI to Rank and Filter Edges in Graphs of Words 1 Introduction 2 Related Work 3 Proposed Approach 3.1 Text Pre-processing 3.2 Graph Modeling 3.3 Graph Representation Learning 3.4 Text CNN for Classification 4 Experiment Setup 4.1 Implementation Details 4.2 Datasets 4.3 Evaluation 5 Results and Discussion 6 Conclusion and Future Work References Development and Research of Quantum Models for Image Conversion 1 Introduction 2 Mathematical Apparatus 3 Quantum Image Processing Theory 3.1 Quantum Image Manipulation 3.2 Quantum Image Transformation 4 Implementing a Quantum Computing Simulation Environment 5 Comparison of Image Transformation Models and Their Classification 6 Conclusions References Selecting the Optimal Configuration of Swarm Algorithms for an NLP Task 1 Introduction 2 The Phrasing Task 2.1 Review of Relevant Work 2.2 Description of the Attraction–Repulsion Phrasing Model 3 Essentials of the Swarm Algorithm 3.1 Description of Particle Swarms Studied 3.2 Swarm Configurations 4 Evaluating the Phrasing Accuracy 5 Analysing the Models Generated During Phrasing 6 Experiments on Swarm Parameters 6.1 Introductory Material 6.2 Comparing the Swarm Topologies 6.3 Comparing the Particle Swarm Variants 6.4 Determining the Statistical Significance of Results 6.5 Convergence of Homogeneous vs. Heterogeneous Swarms 7 Theoretical Insights on Swarm Complexity 8 Conclusions References Active Learning Strategies and Convolutional Neural Networks for Mammogram Classification 1 Introduction 2 Background 3 Proposed Methodology 4 Experiments 4.1 Dataset 4.2 Scenarios 5 Results and Discussion 6 Conclusion References Data Mining Exploiting Time Dynamics for One-Class and Open-Set Anomaly Detection 1 Introduction 2 Problem Definition 3 Log-to-Temporal-Image Anomaly Detection 3.1 Log-to-Temporal Image Transformation 3.2 Class Modelling 3.3 Architecture of the Autoencoders 4 Regression 5 Open-Set Classification 6 Experimental Results 6.1 Anomaly Detection Results 6.2 Open-Set Classification Results 6.3 Pepper Use Case 7 Conclusions References cgSpan: Pattern Mining in Conceptual Graphs 1 Introduction 2 State of the Art 2.1 Conceptual Graphs 2.2 Subgraph Mining 3 cgSpan: A CG Frequent Pattern Mining Algorithm 3.1 Overview 3.2 Exploiting Relation Arity and Neighborhood Nodes 3.3 Exploiting Signatures 3.4 Exploiting Rules 4 Experimental Study 4.1 Data Generation 4.2 Criteria 4.3 Experimental Results 5 Conclusion and Future Works References Constrained Clustering Problems: New Optimization Algorithms 1 Introduction 2 Related Work 3 Clustering Algorithms: A Mathematical Formulation 3.1 Classical K-Means Algorithm 3.2 Constrained Clustering Algorithm 4 Theoretical Analysis 5 Performance Evaluation 5.1 Datasets Description 5.2 Clustering Optimization Evaluation 6 Open Research Challenge 7 Conclusion and Future Work References Robustness of Supervised Learning Based on Combined Centroids 1 Introduction 2 Methodology 2.1 Joint Localization of the Mouth and Eyes 2.2 Optimization for the Joint Localization 2.3 An Approach Based on the Robust Correlation Coefficient 3 Experiments over a Database of Facial Images 3.1 Initial Results 3.2 Modified Test Dataset 4 Conclusions References Cognitive Consistency Models Applied to Data Clustering 1 Introduction 2 Related Work 3 Cognitive Dissonance Clustering Algorithm 4 Results and Discussion 5 Conclusion References Interactive Process Drift Detection Framework 1 Process Mining in Evolving Environments 2 Process Drift 2.1 Process Drift Detection Tools 3 Interactive Process Drift Detection Framework 3.1 Windowing Strategy 3.2 Process Discovery 3.3 Model-to-Model Comparison 3.4 Evaluation 4 Results 5 Conclusion References Mining of High-Utility Patterns in Big IoT Databases 1 Introduction 2 Problem Statement 2.1 Downward Closure Property 3 Proposed Uncertain High Utility Pattern Mining Algorithms 3.1 The Node's Expression for Itemset 4 Experimental Evaluation 4.1 Data Information and Preparation 5 Conclusion References Dynamic Ensemble Selection for Imbalanced Data Stream Classification with Limited Label Access 1 Introduction 2 Methods 3 Experimental Evaluation 4 Conclusions References Various Problems of Ariticial Intelligence Applying and Comparing Policy Gradient Methods to Multi-echelon Supply Chains with Uncertain Demands and Lead Times 1 Introduction 2 Problem Formulation 3 Experimental Methodology 4 Experiments and Results 4.1 First-Phase Experiments 4.2 Second-Phase Experiments 5 Conclusions References Formants Analysis of L2 Arabic Short Vowels: The Impact of Gender and Foreign Accent 1 Introduction 2 Speakers and Speech Material 3 Measurement 4 Results 4.1 L1/L2 Formant Analysis 4.2 Gender L1/L2 Formant Analysis 5 Conclusion References Cluster Analysis of Co-occurring Human Personality Traits and Depression 1 Introduction 2 Method Description 3 Research Description 3.1 Results Description 3.2 Conclusions 4 Final Remarks References Polynomial Algorithm for Solving Cross-matching Puzzles 1 Introduction 2 Problem Description 3 Proposed Solution 4 Conclusion References Credit Risk Assessment in the Banking Sector Based on Neural Network Analysis 1 Introduction 2 Credit Risk Assessment in the Banking Sector 3 Analysis of the Results Obtained 4 Conclusion References Neural Network Model for the Multiple Factor Analysis of Economic Efficiency of an Enterprise 1 Introduction 2 Objective Setting. Neural Network Model for the Multiple Factor Analysis of Economic Efficiency of an Enterprise 3 Development of a Neural Network Model for the Multiple Factor Analysis of Economic Efficiency Based on a Case Study of Rosatom 4 Conclusion References A Study of Direct and Indirect Encoding in Phenotype-Genotype Relationships 1 Introduction to Phenotype and Genotype Mappings 1.1 Direct Encodings for ANNs 1.2 Indirect Encodings for ANNs 1.3 The Evolutionary Search Space of Indirect Encodings 2 Indirect Encoding 3 Summary of Research on Indirect Encoding 3.1 Basic Processing of HyperNEAT 4 Experiment 4.1 Experimental Parameters 4.2 Results 5 Conclusion References Study of the Influences of Stimuli Characteristics in the Implementation of Steady State Visual Evoked Potentials Based Brain Computer Interface Systems 1 Introduction 2 State of Art 3 Methods 3.1 Environment for the Experimentation 3.2 Description of Experiments 4 Results 4.1 Frequency of the Stimulus 4.2 Source of Stimulus 4.3 Color and Illuminance of the Stimulus 4.4 Diffuse (Frosted) Light 4.5 Size of the Stimulus 4.6 Number of Simultaneous Stimuli 5 Discussion 6 Conclusions References Promises and Challenges of Reinforcement Learning Applications in Motion Planning of Automated Vehicles 1 Introduction 1.1 Our Work 2 Promises 3 Challenges 3.1 Training Off-Line from the Fixed Logs of an External Behavior Policy 3.2 Learning on the Real System from Limited Samples 3.3 High-Dimensional Continuous State and Action Spaces 3.4 Satisfying Safety Constraints 3.5 Partial Observability and Non-stationarity 3.6 Unspecified and Multi-objective Reward Functions 3.7 Explainability 4 Summary References The Usage of Possibility Degree in the Multi-criteria Decision-Analysis Problems 1 Introduction 2 Preliminries 3 Methods 3.1 The COMET Method 3.2 Similarity Coefficients 4 The Proposed Approach 5 Comparative Study Case 6 Conclusions References Ant-Based Hyper-Heuristics for the Movie Scene Scheduling Problem 1 Introduction 2 Background 3 Movie Scene Scheduling Problem 3.1 Problem Definition 3.2 Problem Extensions 4 Fast Ant Hyper-Heuristic 4.1 Hyper-Heuristic Adaptations 4.2 Heuristic Search Procedure 4.3 Pheromone Accumulation Vector 4.4 Ant Updates 4.5 Learning Framework and Hybridisation 5 Experimental Setup 5.1 Dataset Generation 5.2 Heuristics 5.3 Parameters 5.4 Experimental Process 5.5 Assessment Metrics 5.6 Technical Specifications 6 Results and Discussion 6.1 Discussion and Analysis 7 Conclusion References AI Alignment of Disaster Resilience Management Support Systems 1 Introduction 1.1 The Resilience Management Support System Aims and Functionalities 1.2 The Aims and the Structure of this Paper 2 Related Work 3 AI Alignment and the Software Architecture of DRMSS 4 AI Evolution Modelling Tool 4.1 The AIEM Implementation 4.2 The Implementation of the DRMSS Operational Modules 5 Summary and Conclusions References A Generative Design Method Based on a Graph Transformation System 1 Introduction 2 Generating Design Drawings 3 CP-Graph Representation of Designed Objects 4 CP-Graph Transformation Rule Inference 5 Automatic Generation of New Deigns 6 Conclusion References Time Dependent Fuel Optimal Satellite Formation Reconfiguration Using Quantum Particle Swarm Optimization 1 Introduction 2 Lambert Theorem 3 QPSO 3.1 QPSO Algorithm 4 Problem Formulation 4.1 Problem Statement 5 Results References The Efficiency of the Stock Exchange - The Case of Stock Indices of IT Companies 1 Introduction 2 Algorithmic Trading 3 Efficient Market Hypotheses, Their Implications and Verification Methods 4 Technical Analysis Methods Used to Create Investment Strategies 5 Methodology of the Study and Results Obtained 6 Conclusions References Simultaneous Contextualization and Interpretation with Keyword Awareness 1 Introduction 2 Related Work 2.1 Multi-sense Embedding 2.2 Dialogue-Context Estimation 2.3 SCAIN 3 SCAIN/KE 3.1 Contextualization 3.2 Interpretation 3.3 Resampling 3.4 Case Study on SCAIN/KE 4 Evaluation 4.1 Method 4.2 Results 5 Conclusions References Aiding Long-Term Investment Decisions with XGBoost Machine Learning Model 1 Introduction 2 Literature Overview 3 Problem Formulation 4 Data Overview 5 Model Outline 6 Train and Test Sample 7 ChangePoints Model 7.1 Contradicting Labels Issue 7.2 Imbalanced Dataset Issue 7.3 Quality Evaluation Issue 7.4 Changepoints XGBoost Realization 7.5 Selecting Model Hyperparameters 8 TrendOrFlat Model 8.1 Dataset Overview 8.2 TrendOrFlat Performance Overview 9 Pipeline Results 10 Conclusion References Bioinformatics, Biometrics and Medical Applications A New Statistical Iterative Reconstruction Algorithm for a CT Scanner with Flying Focal Spot 1 Introduction 2 Scanner Geometry 3 Reconstruction Algorithm 4 Experimental Results 5 Conclusion References A New Multi-filter Framework with Statistical Dense SIFT Descriptor for Spoofing Detection in Fingerprint Authentication Systems 1 Introduction 2 Dense SIFT Fundamentals 3 Proposed Multi-filter Framework and Statistical Dense SIFT for Liveness Detection in Fingerprints 3.1 Multi-filter Framework 3.2 Statistical Dense SIFT 3.3 Proposed Instance 4 Experiments and Results 5 Conclusion References Application of a Neural Network to Generate the Hash Code for a Device Fingerprint 1 Introduction 2 Device Fingerprint 3 Autoencoder 4 Experimental Research Study 5 Conclusions References Fingerprint Device Parameter Stability Analysis 1 Introduction 2 Definition of the Device Fingerprint 3 The Research and Its Results 4 Conclusion References RNA Folding Codes Optimization Using the Intel SDK for OpenCL 1 Introduction 2 Nussinov's RNA Folding Algorithm 3 Related Work 4 The Intel OpenCL Nussinov RNA Folding Algorithm Implementation 5 Code Generator 6 Experimental Study 7 Conclusion References A Comparison of Machine Learning Techniques for Diagnosing Multiple Myeloma 1 Introduction 2 Machine Learning for Multiple Myeloma Diagnosis 3 Diagnosing Multiple Myeloma 4 Machine Learning Approaches 4.1 Artificial Neural Network (ANN) 4.2 Convolutional Neural Network (CNN) 4.3 Random Forest (RF) 4.4 Support Vector Machine (SVM) 5 Experimental Setup 5.1 ANN Parameters 5.2 CNN Parameters 5.3 RF Parameters 5.4 SVM Parameters 5.5 Statistical Tests 6 Results and Discussion 7 Conclusion References Fuzzy Granulation Approach to Face Recognition 1 Introduction 2 Face Description by Use of Fuzzy Sets 3 Type-1 and Type-2 Fuzzy Sets for Face Description 4 Fuzzy Granulation Approach to Face Description 5 Fuzzy Relations in Face Description 6 Fuzzy Relational Rules 7 Face Recognition Based on Linguistic Description 8 Conclusions and Final Remarks References Dynamic Signature Vertical Partitioning Using Selected Population-Based Algorithms 1 Introduction 2 Population-Based Approach for Vertical Signature Partitioning 3 Simulations 4 Conclusions References Author Index