Soft Computing for Problem Solving: Proceedings of the SocProS 2022

Description

This book provides an insight into the 11th International Conference on Soft Computing for Problem Solving (SocProS 2022). This international conference is a joint technical collaboration of the Soft Computing Research Society and the Indian Institute of Technology Mandi. This book presents the latest achievements and innovations in the interdisciplinary areas of Soft Computing, Machine Learning, and Data Science. It brings together the researchers, engineers, and practitioners to discuss thought-provoking developments and challenges, in order to select potential future directions. It covers original research papers in the areas including but not limited to algorithms (artificial neural network, deep learning, statistical methods, genetic algorithm, and particle swarm optimization) and applications (data mining and clustering, computer vision, medical and healthcare, finance, data envelopment analysis, business, and forecasting applications). This book is beneficial for young as well as experienced researchers dealing across complex and intricate real-world problems for which finding a solution by traditional methods is a difficult task. Preface Contents Editors and Contributors Benchmarking State-of-the-Art Methodologies for Optic Disc Segmentation 1 Introduction 2 Dataset 3 Methodology 3.1 Deep Learning Techniques (CNN-Based) 3.2 Adversarial Deep Learning Techniques 4 Results 5 Conclusion References Automated Student Emotion Analysis During Online Classes Using Convolutional Neural Network 1 Introduction 2 Related Works 3 Proposed Scheme 3.1 Dataset 3.2 Data Preprocessing 3.3 Feature Extraction and Emotion Classification Using CNN 4 Experimental Results and Analysis 5 Conclusion References Transfer Learning-Based Malware Classification 1 Introduction 2 Related Work 3 Preliminary 3.1 AlexNet 3.2 Transfer Learning 4 Proposed Work 4.1 Generating Grayscale Images from Malware Samples 4.2 Augmenting the Dataset 4.3 Feature Extraction Using AlexNet 4.4 Sorting the Malware Samples into the Appropriate Families 5 Datasets 6 Experimental Analysis 7 Conclusion References A Study on Metric-Based and Initialization-Based Methods for Few-Shot Image Classification 1 Introduction 2 Background 3 Comparison of Few-Shot Learning Papers 3.1 Distance Metric-Based Learning Methods 3.2 Initialization-Based Methods 4 Experimental Results 5 Conclusion References A Fast and Efficient Methods for Eye Pre-processing and DR Level Detection 1 Introduction 2 Related Work 3 Dataset Description 4 Retina Image Pre-processing 4.1 Why Pre-processing is Required 4.2 The Methodology Used to Pre-process and Implementation 4.3 The Algorithm Used to Pre-process an Input Image 5 Proposed Neural Network Architecture 5.1 Batch Normalization 5.2 Activation Function 5.3 Average Pooling 5.4 Program Flowchart 6 Model Training 7 Conclusion References A Deep Neural Model CNN-LSTM Network for Automated Sleep Staging Based on a Single-Channel EEG Signal 1 Introduction 2 Literature Survey 3 Methodology 3.1 Dataset Used 3.2 Data Preprocessing 3.3 Proposed Deep Neural Network Based on CNN-LSTM 3.4 Model Specification 3.5 Evaluation Methodology 4 Experimental Results and Discussion 5 Discussion 6 Conclusion References An Ensemble Model for Gait Classification in Children and Adolescent with Cerebral Palsy: A Low-Cost Approach 1 Introduction 2 Related Work 3 Methods 3.1 Participants 3.2 Experimental Setup and Data Acquisition 3.3 Data Analysis 4 Results and Discussion 5 Conclusion References Imbalanced Learning of Regular Grammar for DFA Extraction from LSTM Architecture 1 Introduction 2 Related Work 3 Problem Definition 3.1 Tomita Grammar 3.2 Extended Tomita Grammar 3.3 Imbalancing 4 The Proposed Methodology 5 Datasets and Preprocessing 6 Results and Discussion 6.1 Experimental Setup 6.2 Results 6.3 Discussion 7 Conclusion References Medical Prescription Label Reading Using Computer Vision and Deep Learning 1 Introduction 2 Motivation 3 Related Work 4 Design of the Proposed Work 4.1 Data Collection 4.2 Preprocessing 4.3 Training of Data Using Deep Learning 4.4 Evaluation 5 Experimental Results 6 Conclusion and Enhancements References Autoencoder-Based Deep Neural Architecture for Epileptic Seizures Classification 1 Introduction 2 Dataset Description 3 Proposed Approach 3.1 Structure of Autoencoder-Based LSTM 3.2 1D CNN Structure 3.3 Proposed Architecture 4 Model Evaluation and Results 4.1 Binary Classification Task 4.2 Experimental Results and Discussion 5 Conclusions and Future Work References Stock Market Prediction Using Deep Learning Techniques for Short and Long Horizon 1 Introduction 2 Related Work 3 Methodology 4 Experiment 4.1 Data Description 4.2 Assessment Metrics 4.3 Experimental Setup 5 Results and Discussion 6 Conclusion References Improved CNN Model for Breast Cancer Classification 1 Introduction 2 Related Works 3 Proposed Method 3.1 Network Architecture 3.2 Heterogeneous Convolution Module 3.3 Data Preprocessing 4 Results and Analysis 4.1 Experimental Environment 4.2 Training Strategy 4.3 Evaluation Criteria 4.4 Experimental Results and Analysis 5 Conclusions References Performance Assessment of Normalization in CNN with Retinal Image Segmentation 1 Introduction 2 Literature Review 3 Problem Definition: Research Questions 4 The Proposed Methodology 4.1 CNN Architecture 4.2 Normalization Techniques 5 Results and Discussion 5.1 Datasets and Preprocessing 5.2 Experimental Setup 5.3 Results 5.4 Discussion 6 Conclusion References A Novel Multi-day Ahead Index Price Forecast Using Multi-output-Based Deep Learning System 1 Introduction 2 Related Works 3 Methodology 3.1 Artificial Neural Networks (ANNs) 3.2 Long Short-term Memory Networks (LSTMs) 3.3 Proposed Hybrid Model (CNN-LSTM) 4 Data Pre-processing and Feature Engineering 4.1 Dataset 4.2 Technical Indicators 4.3 Random Forest-Based Feature Importance 4.4 Scaling the Training Set 5 Proposed Price Forecasting Framework 5.1 Model Calibration 5.2 Model Evaluation 6 Experimental Results 6.1 Generalizability 7 Conclusion and Future Work References Automatic Retinal Vessel Segmentation Using BTLBO 1 Introduction 2 Related Work 2.1 Retinal Vessel Segmentation 2.2 Neural Architecture Search (NAS) 2.3 BTLBO 3 Methodology 3.1 U-net 3.2 Search Space and Encoding 3.3 BTLBO 4 Experiments and Results 4.1 Dataset 4.2 Metrics 4.3 Results 5 Conclusion References Exploring the Relationship Between Learning Rate, Batch Size, and Epochs in Deep Learning: An Experimental Study 1 Introduction 2 Proposed Methodology 3 Datasets 4 Results 4.1 Using the Proposed Synergy Between Learning Rate, Batch Size, and Epochs 4.2 Introducing Some Randomness in Learning Rate 4.3 Experiments on Other Datasets 5 Conclusion References Encoder–Decoder (LSTM-LSTM) Network-Based Prediction Model for Trend Forecasting in Currency Market 1 Introduction 2 Methodology 2.1 LSTM Block 2.2 Encoder–Decoder Network 2.3 Encoder Layer 2.4 Decoder Layer 2.5 Combination of Encoder and Decoder Architectures 3 Model Formulation and Implementation 4 Description of Experimental Data 5 Performance Measure and Implementation of Prediction Model 5.1 Recall 5.2 Precision 5.3 upper F 1F1-Score 6 Result and Discussion 7 Conclusion References Histopathological Nuclei Segmentation Using Spatial Kernelized Fuzzy Clustering Approach 1 Introduction 2 Related Work 3 Background 3.1 Fuzzy C-Means Clustering 3.2 Kernel Methods and Functions 4 Proposed Methodology: Spatial Circular Kernel Based Fuzzy C-Means Clustering Algorithm (SCKFCM) 5 Results 5.1 Dataset 5.2 Quantitative Evaluation Metrics 5.3 Performance Evaluation 6 Conclusion References Tree Detection from Urban Developed Areas in High-Resolution Satellite Images 1 Introduction 2 A Designed Framework for Tree Region Detection Using Thresholding Approach 2.1 Automatic Thresholding-Based Tree Region Detection in the Satellite Images 2.2 Region Growing-Based Tree Region Detection in the Satellite Images 3 Results and Discussion 3.1 Accuracy Assessment 4 Results and Discussion References Emotional Information-Based Hybrid Recommendation System 1 Introduction 2 Related Work 3 Proposed Model 3.1 Content-Based Method 3.2 Collaborative Filtering Method 3.3 Methods for Evaluating the Models 4 Experimentation and Results 4.1 Setup 4.2 Dataset Used 4.3 Quantitative Analysis 4.4 Qualitative Analysis 4.5 Result Comparison 4.6 Future Insights 5 Conclusion References A Novel Approach for Malicious Intrusion Detection Using Ensemble Feature Selection Method 1 Introduction 2 Related Work 3 Proposed Work and Implementation 3.1 Proposed Ensemble-Based Feature Selection 3.2 Training Process 3.3 Testing Process 4 Analysis and Discussion 4.1 Feature Selection Method Based Results 4.2 Classifier-Based Results on EFS Applied Dataset 5 Conclusion References Automatic Criminal Recidivism Risk Estimation in Recidivist Using Classification and Ensemble Techniques 1 Introduction 2 Data and Methods 2.1 Study Subject Selection 2.2 Data Acquisition 2.3 Data Preprocessing 2.4 Data Quantification and Transformation 2.5 Classification 2.6 Proposed Methodology 3 Results 4 Conclusion References Assessing Imbalanced Datasets in Binary Classifiers 1 Introduction 2 Related Work 3 The Methodology 4 Datasets and Preprocessing 5 Experimental Results 5.1 Relation Between Imbalance Ratio and Accuracy Rate 6 Conclusion References A Hybrid Machine Learning Approach for Multistep Ahead Future Price Forecasting 1 Introduction 2 Methodology/Mathematical Background 2.1 Support Vector Regression 2.2 Least Square Support Vector Regression (LS-SVR) 2.3 Proximal Support Vector Regression (PSVR) 2.4 Feature Dimensionality Reduction 2.5 Kernel Principal Component Analysis 3 Proposed Hybrid Approach 3.1 Input Feature 3.2 Multistep Ahead Forecast Price 3.3 Proposed Hybrid Models 4 Results and Discussion 4.1 Datasets 4.2 Performance Evaluation Criteria 4.3 Result Analysis 5 Conclusion References Soft Computing Approach for Student Dropouts in Education System 1 Introduction 2 Preliminaries 2.1 Support Vector Machine 2.2 Naïve Bayes 2.3 N-Gram 3 Related Work 4 Methodology 4.1 Collection of Data 4.2 Preprocessing of the Data 4.3 Categorizing the Data 4.4 Extraction of Data 4.5 Evaluation Report 5 About Dataset 6 Proposed Model 7 Result and Discussion 8 Conclusion and Future Scope References Machine Learning-Based Hybrid Models for Trend Forecasting in Financial Instruments 1 Introduction 2 Methodology 2.1 Classification Models 2.2 Feature Selection Methods 3 Proposed Hybrid Methods 3.1 Input 3.2 Hybrid Models 3.3 Training and Parameter Selection 4 Experiment and Discussion 4.1 Data Description 4.2 Performance Measures 4.3 Results and Discussion 5 Conclusion References Support Vector Regression-Based Hybrid Models for Multi-day Ahead Forecasting of Cryptocurrency 1 Introduction 2 Methodology 2.1 Forecasting Methods 2.2 Feature Selection 3 Proposed Forecasting Model 3.1 Cryptocurrency 3.2 Input Features 3.3 System Architect 3.4 Multi-step Ahead Forecasting Strategies 4 Experiments and Discussion 4.1 Dataset Description 4.2 Performance Analysis 4.3 Parameter Selection 4.4 Implementation of Forecasting Model 4.5 Results and Discussion 5 Conclusion References Image Segmentation Using Structural SVM and Core Vector Machines 1 Introduction 2 Methodology 2.1 Structural Support Vector Machines (SSVM) 2.2 Core Vector Machines (CVM 3 Results and Discussion 3.1 Segmentation Using SSVM and CVM 3.2 Experimental Dataset Description 3.3 Performance Measure 3.4 Implementation of Prediction Model 3.5 Experimental Results 4 Conclusion and Scope References Identification of Performance Contributing Features of Technology-Based Startups Using a Hybrid Framework 1 Introduction 2 Proposed Framework 3 Results 4 Conclusion References Fraud Detection Model Using Semi-supervised Learning 1 Introduction 2 Methodology 2.1 Working of Laplacian Models 2.2 Importance of Unlabeled Data in SSL 2.3 SSL Procedure 2.4 Assumptions in SSL 2.5 Why Manifolds? 2.6 Manifold Regularization 2.7 Laplacian SVM 2.8 Mathematical Formulation 3 Proposed Fraud Detection Model 3.1 Experimental Setup 3.2 Results 4 Conclusion References A Modified Lévy Flight Grey Wolf Optimizer Feature Selection Approach to Breast Cancer Dataset 1 Introduction 2 Literature Review 3 Materials and Methods 3.1 Details on Dataset 3.2 Grey Wolf Optimization (GWO) 3.3 Grey Wolf Based on Lévy Flight Feature Selection Method 4 Experimental Results 4.1 Performance Evaluation 4.2 Relevant Feature Selected 5 Conclusion References Feature Selection Using Hybrid Black Hole Genetic Algorithm in Multi-label Datasets 1 Introduction 1.1 Multi-label Classification 2 Related Works 3 Proposed Method 3.1 Standalone Binary Black Hole Algorithm (SBH) 3.2 Improved Hybrid Black Hole Genetic Algorithm for Multi-label Feature Selection 3.3 Datasets 3.4 Simulation Setup 4 Experimental Results 4.1 Dataset-I 4.2 Dataset-II 4.3 Computational Complexity 5 Conclusion References Design and Analysis of Composite Leaf Spring Suspension System by Using Particle Swarm Optimization Technique 1 Introduction 2 Literature Review 3 Problem Statement 4 Conventional Leaf Spring 5 Composite Leaf Spring 5.1 Objective Function 5.2 Design Variables 5.3 Design Parameters 5.4 Design Constraints 6 Particle Swarm Optimization 6.1 Velocity Clamping 7 Algorithm 8 Result 9 Conclusion References Superpixel Image Clustering Using Particle Swarm Optimizer for Nucleus Segmentation 1 Introduction 2 Methodology 2.1 Superpixel Generating Techniques 2.2 SLIC (Simple Linear Iterative Clustering) Algorithm for Making Superpixels 2.3 Objective Function of Superpixel Image-Based Segmentation 2.4 Particle Swarm Optimization (PSO) 3 Results and Dıscussıon 3.1 Results and Discussion of Kidney Renal Cell Images 4 Conclusıon References Whale Optimization-Based Task Offloading Technique in Integrated Cloud-Fog Environment 1 Introduction 1.1 Task Offloading 2 Literature Review 3 Whale Optimization Algorithm 4 Architecture of Integrated Cloud-Fog Environment 4.1 IoT Layer 4.2 Fog Layer 4.3 Cloud Layer 5 Result and Discussion 5.1 Experimental Setup 5.2 Experimental Analysis 6 Conclusion References Solution to the Unconstrained Portfolio Optimisation Problem Using a Genetic Algorithm 1 Introduction 2 Multi-objective Optimisation Problems 3 Portfolio Optimization Problem 4 Genetic Algorithms 4.1 Encoding 4.2 Fitness Evaluation 4.3 Selection 4.4 Crossover 4.5 Mutation 5 Performance Metrics 5.1 Set Coverage Metric 5.2 Generational Distance 5.3 Maximum Pareto-Optimal Front Error 5.4 Spacing 5.5 Spread 5.6 Maximum Spread 6 Result and Analysis 6.1 Set Coverage Metric 6.2 Generational Distance and MFE 6.3 Spacing 6.4 Spread and Maximum Spread 7 Conclusion 7.1 Future Scope References Task Scheduling and Energy-Aware Workflow in the Cloud Through Hybrid Optimization Techniques 1 Introduction 2 Related Work 3 Conclusion References A Hyper-Heuristic Method for the Traveling Repairman Problem with Profits 1 Introduction 2 Overview of Hyper-Heuristic 3 Proposed Hyper-Heuristic Method 3.1 Generation of the Initial Solution 3.2 Low-Level Heuristics 3.3 Proposed Algorithm 3.4 Complexity Analysis of HH-GREEDY 4 Computational Results 5 Conclusions References Economic Dispatch Using Adapted Particle Swarm Optimization 1 Introduction 2 Economic Dispatch (ED) Problem 3 Proposed Adapted Particle Swarm Optimization (aPSO) 3.1 Standard PSO 3.2 Proposed aPSO 4 Application and Results 5 Conclusion and Future Work References A Mathematical Model to Minimize the Total Cultivation Cost of Sugarcane 1 Introduction 1.1 A Sugarcane Supply Chain 2 Literature Review 3 Problem Formulation 3.1 Mathematical Model 4 Methodology 4.1 Data Collection 4.2 Differential Evolution 4.3 Particle Swarm Optimization 4.4 Parameter Setting 4.5 System Configuration 5 Results and Discussion 5.1 Comparison of Expenditure (Actual v/s PSO and DE) 6 Conclusion and Future Directions References Genetically Optimized PID Controller for a Novel Corn Dryer 1 Introduction 2 Literature Survey 3 Problem Statement 4 Proposed Technique 4.1 PID Function: Part 1 4.2 Part 2 5 Result Analysis 6 Conclusion References Minimization of Molecular Potential Energy Function Using Laplacian Salp Swarm Algorithm (LX-SSA) 1 Introduction 2 Molecular Potential Energy Problem 3 Laplacian Salp Swarm Algorithm (LX-SSA) 3.1 Computational Steps 4 Performance Evaluation Criteria 5 Numerical Results 6 Conclusion References Performance Evaluation by SBM DEA Model Under Fuzzy Environments Using Expected Credits 1 Introduction 2 Preliminaries 2.1 Slacks Based Measure DEA Model 2.2 Fuzzy Numbers 2.3 Fuzzy SBM DEA Model 3 Expected Credits 4 Numerical Illustration 4.1 Inputs and Output 5 Conclusion References Measuring Efficiency of Hotels and Restaurants Using Recyclable Input and Outputs 1 Introduction 2 Literature Review 3 Data Envelopment Analysis 3.1 DEA Model 4 Research Design 4.1 Selection of DMUs 4.2 Data and Variables 5 Results and Discussions 5.1 Overall Performance of H&R 6 Post-DEA Analysis 6.1 Recyclable Input–output Analysis 7 Conclusion References Efficiency Assessment of an Institute Through Parallel Network Data Envelopment Analysis 1 Introduction 2 Methodology 2.1 CCR Model 2.2 Parallel Network DEA 2.3 Mathematical Model of Parallel NDEA 3 Problem Structure 4 Results and Discussion 4.1 Assessment Through the Conventional DEA Model 4.2 Assessment Through Parallel Network DEA Model 5 Conclusion References Efficiency Measurement at Major Ports of India During the Years 2013–14 to 2018–19: A Comparison of Results Obtained from DEA Model and DEA with Shannon Entropy Technique 1 Introduction 2 Literature Review 2.1 Studies Covering International Ports 2.2 Studies Covering Indian Ports 3 Research Methodology 3.1 Data Envelopment Analysis 3.2 Integration of Shannon’s Entropy with DEA 3.3 Data Collection for Performance Measurement 4 Analysis of Results 5 Findings, Conclusions, and Scope for Further Research References Ranking of Efficient DMUs Using Super-Efficiency Inverse DEA Model 1 Introduction 2 Research Methodology 2.1 CCR Model 2.2 Inverse DEA Model 2.3 Super-Efficiency DEA Model 2.4 Super-Efficiency Inverse DEA Model 2.5 Single-Objective IDEA Model 3 Numerical Illustration 3.1 Data and Parameters Collection: 3.2 Empirical Results 4 Conclusion References Data Encryption in Fog Computing Using Hybrid Cryptography with Integrity Check 1 Introduction 1.1 Data Security Issues in Fog Computing 2 Related Works 3 Proposed System 3.1 Sender’s Architecture (Encryption) 3.2 Receiver’s Architecture (Decryption) 3.3 Simulation Settings 3.4 Data and Performance Metric 4 Results and Analysis 4.1 Results Based on Encryption and Throughput-Encryption 4.2 Results Based on Decryption and Throughput-Decryption 4.3 Comparative Analysis of Results Obtained 5 Conclusion References Reducing Grid Dependency and Operating Cost of Micro Grids with Effective Coordination of Renewable and Electric Vehicle’s Storage 1 Introduction 2 EV Mobility Modeling 2.1 Electric Vehicle Mobility Data 2.2 Electric Vehicle Laxity 2.3 Zones of Energy Need 2.4 Energy Distribution 3 Electric Vehicle Usages Probability 3.1 Transition Probability ‘CS → CS’, p11t 3.2 Transition Probability ‘CH → DC’, P12t 3.3 Transition Probability ‘CH → IDL’, P13t 3.4 Transition Probability ‘DC → CH’, p21t 3.5 Transition Probability ‘DC → DC’, p22t 3.6 Transition Probability ‘DC → IDL’, p23tp23t 4 Electric Vehicle Prioritization 4.1 ANFIS Prioritization Procedure 4.2 ANFIS Training Data 5 Results and Analysis 6 Conclusion References A Review Survey of the Algorithms Used for the Blockchain Technology 1 Introduction 2 Literature Review 3 Blockchain 3.1 Blockchain Structure 3.2 Working of Blockchain 3.3 Types of Blockchain 3.4 Characteristics of Blockchain 4 Algorithms Used in Blockchain 4.1 Cryptography Algorithms 4.2 Peer-To-Peer Network Protocol 4.3 Zero-Knowledge Proofs 4.4 Consensus Algorithms 5 Future Trends 6 Conclusion References Relay Coordination of OCR and GFR for Wind Connected Transformer Protection in Distribution System Using ETAP 1 Introduction 2 Power System Faults 2.1 Symmetrical Faults 2.2 Unsymmetrical Faults 2.3 Short Circuit Analysis 3 Power System Protection and Relaying 4 Methodology for Load Flow Analysis 5 Conclusion References Localized Community-Based Node Anomalies in Complex Networks 1 Introduction 2 Related Work 3 Proposed Methodology 3.1 Problem Definition 3.2 Proposed Algorithm 3.3 Mathematical Explanation of Our Proposed Algorithm 4 Results and Discussion 4.1 Network Data Statistics 4.2 Results 4.3 Discussion 5 Conclusion References Time Series Analysis of National Stock Exchange: A Multivariate Data Science Approach 1 Introduction 1.1 Objective 1.2 Background 2 Related Work 3 Methodology 3.1 Finding and Eliminating Missing Values 3.2 Descriptive Analysis 3.3 Multiple Linear Regression (MLR) 3.4 Prediction Analysis 3.5 Rank Correlation 3.6 Multicollinearity: A Potential Problem 3.7 Test of Linearity 3.8 ARIMA 3.9 MLR Versus ARIMA 4 Results and Discussion 4.1 Descriptive Analysis 4.2 Regression Analysis: (MLR) 4.3 Prediction Analysis 4.4 Rank Correlation 4.5 Multicollinearity 4.6 ARIMA 4.7 Test of Linearity 4.8 Validation 5 Conclusion 6 Future Scope References A TOPSIS Method Based on Entropy Measure for qq-Rung Orthopair Fuzzy Sets and Its Application in MADM 1 Introduction 2 Preliminaries 3 A New Constructive Q-ROF Entropy 4 A TOPSIS Approach for MADM Based on the Proposed Entropy Measure of qq-ROFNs 5 Illustrative Example 6 Conclusion References A Novel Score Function for Picture Fuzzy Numbers and Its Based Entropy Method to Multiple Attribute Decision-Making 1 Introduction 2 Preliminaries 3 Shortcomings of the Existing Score Functions Under PFS Environment 4 A Novel Score Function for PFNs 5 Proposed Algorithm for Solving MADM Problem Under PFS Framework 6 Numerical Example 6.1 A Comparative Study with the Existing Methods 7 Conclusion References Author Index