Women in Computational Intelligence: Key Advances and Perspectives on Emerging Topics (Women in Engineering and Science)

This book provides a breadth of innovative and impactful research in the field computational intelligence led by women investigators. Topics include intelligent data analytics, optimization of complex systems, approximation of human reasoning, robotic path planning, and intelligent control systems. These topics touch on many of the technological challenges facing the world today and these solutions by women researcher teams are valuable for their excellence and their non-traditional perspective. As an important part of the Women in Science and Engineering book series, the work highlights the contribution of women leaders in computational intelligence, inspiring women and men, girls, and boys to enter and apply themselves to this exciting multi-disciplinary field. Provides insight into womens contributions to the field of computational intelligence; Presents research from academia, research, and industry on advances, applications, and challenges in computational intelligence; Includes topics such as fuzzy logic, neural networks, and evolutionary computation Preface Contents Amazing Grace – Computer Pioneer Admiral Grace Murray Hopper 1 Introduction 2 Early Years 3 Education and Early Career 4 World War II and Harvard 5 Moving into Industry 6 The First Compiler 7 The US Navy 8 Teaching Career 9 Awards and Honors 10 Lasting Influences References Part I Intelligence XAI: A Natural Application Domain for Fuzzy Set Theory 1 Introduction 2 Computing with Words 2.1 Formal Tools 2.1.1 Fuzzy Sets 2.1.2 Linguistic Variables and Fuzzy Partitions 2.1.3 Fuzzy Quantifiers 2.2 Examples of Methodological Utilisation 2.2.1 Fuzzy Databases 2.2.2 Fuzzy Linguistic Summaries 2.2.3 Extended Linguistic Summaries 2.2.4 Computing with Words Conclusion 3 Fuzzy Approximate Reasoning 3.1 General Principles 3.2 Generalised Modus Ponens 3.2.1 Modus Ponens 3.2.2 Fuzzy Extension 3.3 Other Reasoning Forms 3.3.1 Gradual Reasoning 3.3.2 Analogical Reasoning 3.3.3 Interpolative Reasoning 3.3.4 Fuzzy Reasoning Conclusion 3.4 Examples of Methodological Utilisation 3.4.1 Fuzzy Ontologies 3.4.2 Fuzzy Control 4 Fuzzy Machine Learning 4.1 Fuzzy Rule-Based Systems 4.2 Fuzzy Decision Trees 4.3 Fuzzy Clustering 5 Conclusion References Adaptive Psychological Profiling from Nonverbal Behavior – Why Are Ethics Just Not Enough to Build Trust? 1 Introduction 2 Machine-Based Automated Psychological Profiling 2.1 Nonverbal Behavior 2.2 Automated Psychological Profiling Systems 2.3 Ethical and Legal Implications of Automated Profiling 2.4 The Rise of Ethical Charters 3 Case Study 1: Deception 3.1 Overview of iBorderCtrl 3.2 Privacy and Security by Design 3.3 Project Ethics 3.4 Automated Deception Detection Tool 3.5 Media and Public Discourse 4 Case Study 2: Comprehension 4.1 FATHOM 4.2 COMPASS 5 Empowering the General Public Through Education 6 Conclusions References Conversational Intelligent Tutoring Systems: The State of the Art 1 Introduction 2 Conversational Intelligent Tutoring Systems 2.1 Design Challenges for CITS 3 Oscar CITS 3.1 Automatic Profiling of Learning Styles 3.1.1 Learning Styles Knowledge Engineering 3.1.2 Capturing CITS Behaviour Dataset 3.1.3 Learning Styles Prediction Approaches 4 Hendrix 2.0 CITS 4.1 Profiling Comprehension: Comprehension Assessment and Scoring System (COMPASS) 5 Ethical Use of `AI' in Education 6 Scalability Challenges for CITS 7 Conclusion and Future Directions References Design and Validation of a Mini-Game for Player Motive Profiling 1 Introduction 2 Game Design 2.1 Storyline 2.2 Game Mechanics 2.3 Non-player Characters 2.4 Gameplay 3 Experimental Validation 3.1 Analysing Play Strategies of Non-player Characters for Assessing Motivation 3.2 Analysing Features of Non-player Characters for Assessing Motivation 3.3 Analysing Individual Play and Social Network Phases for Assessing Motivation 4 Conclusion and Future Work References When AI Meets Digital Pathology Acronyms 1 Evolution of Digital Pathology 2 Data Issues and Preprocessing 3 Multiscale Convolutional Neural Networks for Tumor Detection 4 Cell Detection and Segmentation 4.1 Dilemmas Associated with Annotation Loading and Data Imbalance 4.2 Lessening the Burden of Manual Annotation 5 System Integration and Interoperability References Linguistic Intelligence as a Base for Computing Reasoning 1 Introduction 2 Language as a Tool for Communication 2.1 MLW 2.2 Sounds and Utterances Behavior 2.2.1 Spoken Language and L-Systems 2.2.2 Dragon Curves 2.2.3 Dialogue Content and Its Relation with Energy Distribution 2.2.4 Spoken Language Analyses 3 Conclusions and Future Work References Part II Learning Intrusion Detection: Deep Neural Networks Versus Super Learning 1 Introduction 2 Related Work 3 Approaches 3.1 Deep Neural Networks 3.2 Super Learner 3.3 Gradient Boosting Estimator 3.4 Random Forest Estimator 3.5 XGBoost 4 Data Set and Environment 4.1 Data Set 4.2 Evaluation Measures 4.3 H2O Environment 5 Experiments and Results 5.1 Parameter Setup 5.2 Experiments with DNN 5.3 Experiments with Superlearner 6 Conclusion References Lifelong Learning Machines: Towards Developing Optimisation Systems That Continually Learn 1 Introduction 2 Background and Motivation 2.1 Related Work 3 An Immune-Inspired Approach to Lifelong Learning 3.1 A Brief Primer on the Immune System 3.2 NELLI: Mapping to a Lifelong Learning Optimiser 4 Implementation 4.1 Application: Bin-Packing 4.2 Heuristic Generation 4.3 Network 5 Demonstration 6 Conclusion References Reinforcement Learning Control by Direct Heuristic Dynamic Programming 1 Introduction 2 The direction Heuristic Dynamic Programming (dHDP) as an Actor-Critic Type Reinforcement Learning Controller 2.1 Building Blocks of the dHDP 2.2 Properties of dHDP Learning Controller 3 Applications of the dHDP for Automatic Tuning of Prosthesis Control Parameters 3.1 The Robotic Knee Prosthesis Control Problem and Current Approaches 3.2 The dHDP for Automatic Tuning of Robotic Prosthesis 4 Conclusion References Distributed Machine Learning in Energy Management and Control in Smart Grid 1 Introduction 2 Emerging Trends in Distributed Energy Management and Control in Smart Grids 3 Distributed Control Algorithms in Smart Grids 3.1 Consensus-Based Approach 3.2 Dual Decomposition 3.3 Optimality Condition Decomposition 3.4 Population-Based Distributed Algorithms 4 Energy Management and Control in Grid-Connected Microgrids 4.1 Operation Cost 4.2 Power Losses 4.3 Network Voltage Regulation 4.4 Line Limits 4.5 Power Flow Modeling Constraints 4.6 Reactive Power Limit Constraints 4.7 Reactive Power Ramp-Rate Constraints 5 Proposed Distributed Algorithm for Optimal Control of Smart Grids 5.1 Consensus Algorithm 5.2 Choice of PSO as Basis for the Proposed Distributed CI Algorithm 5.3 Proposed Consensus-Based and Dimension-Distributed PSO-Based Algorithm 6 Simulation Results and Discussion 6.1 Simulation Studies on 30-Node Test System 6.1.1 Case Study 1—Convergence, Adaptability, and Performance Benchmark for the 30-Node Test System 6.1.2 Case Study 2—Performance Benchmark with Centralized Controller for 30-Node Test System 6.2 Simulation Studies on the 119-Node Test System 6.2.1 Case Study 1—Convergence, Adaptability, and Performance Benchmark for 119-Node Test System 6.2.2 Case Study 2—Performance Benchmark with Centralized Controller for 119-Node Test System 7 Conclusions References Part III Modeling Fuzzy Multilayer Perceptrons for Fuzzy Vector Regression 1 Introduction 2 Fuzzy Multilayer Perceptron with Cuckoo Search 3 Experiment Results 3.1 Yacht Hydrodynamics Data Set 3.2 Energy Efficiency Data Set 3.3 Ping River Data Set 4 Conclusion References Generalisation in Genetic Programming for Symbolic Regression: Challenges and Future Directions 1 Genetic Programming for Symbolic Regression 1.1 Symbolic Regression 1.2 Genetic Programming—An Evolutionary Computation Technique 1.2.1 Representation 1.2.2 Initialisation 1.2.3 Evaluation 1.2.4 Selection 1.2.5 Genetic Operators 1.3 GP for Symbolic Regression 2 Generalisation in GPSR 2.1 Concepts Related to Generalisation 2.1.1 Overfitting 2.1.2 Bias–Variance Decomposition 3 Enhancing the Generalisation Ability of GPSR 3.1 Data Sampling and Feature Selection 3.1.1 Data Sampling 3.1.2 Feature Selection for GPSR 3.2 Estimating Generalisation Errors 3.2.1 Validation Methods 3.2.2 Estimating Variance Errors 3.2.3 Model Complexity and Generalisation Error 3.3 Improving Selection and Genetic Operators for Better Generalisation 3.4 Ensemble Learning 4 Conclusions and Future Directions References Neuroevolutionary Models Based on Quantum-Inspired Evolutionary Algorithms 1 Introduction 2 Quantum-Inspired Evolutionary Algorithms 2.1 Quantum-Inspired Evolutionary Algorithm with Real Representation (QIEA-R) 2.2 Quantum-Inspired Evolutionary Algorithm with Binary-Real Representation (QIEA-BR) 2.3 Quantum-Inspired Evolutionary Algorithm with Categorical Representation 3 Neuroevolutionary Models Based on QIEA 3.1 Multi-layer Perceptrons 3.2 Fully Recurrent Neural Networks 3.3 Echo States Networks 3.3.1 Phase 1: Global Parameter Optimisation 3.3.2 Phase 2: Reservoir Optimisation 3.4 Convolutional Neural Networks 4 Case Studies 4.1 Classification: Concept Drift Environment 4.1.1 Classification in Concept Drift Scenarios Using Quantum-Inspired Neuroevolution 4.1.2 Experiments 4.2 System Identification 4.2.1 Robot Arm Results 4.3 Deep Learning with Neural Architecture Search (Q-NAS) 5 Conclusions and Future Work References Weightless Neural Models: An Overview 1 Introduction 2 Weightless Neural Networks 3 Quantum Weightless Networks 4 Conclusions and Future Directions References Part IV Optimization Challenges Applying Dynamic Multi-objective Optimisation Algorithms to Real-World Problems 1 Introduction 2 Background 2.1 Multi-objective Optimisation Problems 2.2 Dynamic Multi-objective Optimisation Problems 2.3 Multi-objective Algorithms 2.3.1 Multi-objective Evolutionary Computation Algorithms 2.3.2 Multi-objective Swarm Intelligence Algorithms 3 Real-World Problems 3.1 Taxonomy of Real-World Dynamic Multi-objective Optimisation Problems 3.2 Characteristics of Real-World Dynamic Multi-objective Optimisation Problems 4 Challenges of Solving Real-World Problems 4.1 Discrete- Versus Continuous-Valued Variables 4.2 Benchmarks 4.3 Performance Measures 4.4 Algorithm Selection 4.5 Decision-Maker 5 Conclusion References Computational Intelligence Methodologies for Multi-objective Optimization and Decision-Making in Autonomous Systems 1 Introduction 2 Decision-Making in Various Disciplines 3 Multi-objective Optimization and Decision-Making 4 Multi-objective Optimization and Decision-Making in Autonomous Systems 4.1 Decision-Making During the Mission 4.1.1 Computational Time for Optimization 4.1.2 Dynamic Multi-objective Optimization 4.1.3 Multi-modality in Optimization 4.2 Decision-Making in Time Critical Missions 4.2.1 Limited Time for Decision-Making 4.2.2 Reducing the Number of Alternatives 5 Conclusions References A Framework-Based Approach for Flexible Evaluation of Swarm-Intelligent Algorithms 1 Introduction 1.1 Relevance of Swarm-Inspired Algorithms 1.2 General Applicability 1.3 Need for Fair and Flexible Evaluation Methodology 1.4 Requirements on the Evaluation Methodology 1.5 Related Work 2 Proposed Methodological Approach 2.1 Mathematical Methods 2.2 Interdisciplinary Methods 2.3 Software Engineering Methods 3 Experiences with Swarm-Inspired Algorithms and Self-Organizing Use Cases 4 Evaluation 4.1 Algorithm Recommendation for Use Cases 4.1.1 Load Balancing 4.1.2 Load Clustering 4.1.3 Information Placement and Retrieval 4.1.4 P2P Streaming 4.1.5 Routing in P2P 4.2 Theoretical Considerations 4.2.1 Convergence 4.3 Advantages of Space-Base Coordination Middleware 5 Conclusion References An Improved Bat Algorithm with a Dimension-Based Best Bat for Numerical Optimization 1 Introduction 2 Related Works 2.1 The Bat Algorithm 2.2 Some BA Variants 3 The Proposed DBBA 3.1 Dimension-Based Best Bat 3.2 The Framework of DBBA 4 Experimental Results and Discussion 4.1 Benchmark Functions and Simulation Environment 4.2 Comparison of DBBA with a Standard BA 4.3 Comparison of the DBBA with Some Related Algorithms 4.4 Analysis on the Ordering Sensitivity 5 Concluding Remarks References Index