Fuzzy Logic Hybrid Extensions of Neural and Optimization Algorithms: Theory and Applications

Preface
Contents
Estimation of the Number of Filters in the Convolution Layers of a Convolutional Neural Network Using a Fuzzy Logic System
	1 Introduction
	2 Literature Review
		2.1 Convolutional Neural Networks
		2.2 GSA
		2.3 FGSA
	3 Proposed Method
	4 Results and Discussion
	5 Conclusions
	References
Optimization of Membership Function Parameters for Fuzzy Controllers in Cruise Control Problem Using the Multi-verse Optimizer
	1 Introduction
	2 Fuzzy Systems
		2.1 Mamdani Model
		2.2 Sugeno Model
	3 Control Systems
	4 Metaheuristics and Multi-verse Optimizer
		4.1 Multi-verse Optimizer
		4.2 Applications of MVO
	5 Test and Results
		5.1 Benchmark Function Test and Results
		5.2 Applications Test and Results
	6 Conclusions
	References
Performance Analysis of a Distributed Steady-State Genetic Algorithm Using Low-Power Computers
	1 Introduction
	2 Distributed Steady-State Genetic Algorithm
		2.1 Application of Distributed Steady-State Genetic Algorithm in the n-Queens Problem
		2.2 Application of Distributed Steady-State Genetic Algorithm in the Travelling Salesman Problem
	3 Master-Slave Low Power Architecture
		3.1 Rationale on Master-Slave Architecture Starting Procedure
		3.2 Function Evaluation Task on Slave-Devices
		3.3 Fail-Safe Algorithm on Master-Device
	4 Computational Results
		4.1 Experimental Setup
		4.2 n-Queens Problem Experimental Arrangement Results
		4.3 Travelling Salesman Problem Results
	5 Conclusions and Future Work
	References
Ensemble Recurrent Neural Networks for Complex Time Series Prediction with Integration Methods
	1 Introduction
	2 Problem Statement and Proposed Method
		2.1 Analyze the Time Series
		2.2 Creation of the Recurrent Neural Network
		2.3 Integration by Average
		2.4 Integration by Weighted Average
		2.5 Integration by Gating Network
		2.6 Type-1 and Type-2 Fuzzy System Integration
		2.7 Generalized Type-2 Fuzzy System
	3 Simulation Results
	4 Conclusions
	References
Genetic Optimization of Ensemble Neural Network Architectures for Prediction of COVID-19 Confirmed and Death Cases
	1 Introduction
	2 Basic Concepts
		2.1 Artificial Neural Networks
		2.2 Nonlinear Autoregressive Neural Network
		2.3 Fuzzy Logic
		2.4 Genetic Algorithms
	3 Proposed Method
	4 Results of the Experiment
		4.1 Genetic Algorithms
	5 Conclusions
	References
Optimization of Modular Neural Networks for the Diagnosis of Cardiovascular Risk
	1 Introduction
	2 Literature Review
		2.1 Flower Pollination Algorithm
		2.2 Bird Swarm Algorithm
		2.3 Blood Pressure and Hypertension
		2.4 Cardiovascular Disease and Heart Age
		2.5 Framingham Heart Study
	3 Proposed Method
	4 Results
	5 Conclusions and Future Work
	References
A Review on the Cuckoo Search Algorithm
	1 Introduction
	2 An Analogy with Nature
		2.1 Cuckoo Search Algorithm
		2.2 Algorithm Rules
		2.3 Levy Flights
		2.4 Mathematical Formulas
		2.5 Flowchart CS
	3 Implementation of Levy Flights in Other Algorithms
	4 Variants of the Cuckoo Search Algorithm
	5 Applications
	6 Conclusions
	References
An Improved Convolutional Neural Network Based on a Parameter Modification of the Convolution Layer
	1 Introduction
	2 Background and Basic Concepts
		2.1 Convolutional Neural Network Concepts
		2.2 Edge Detectors
		2.3 Sobel Operator
		2.4 Prewitt Operator
		2.5 Laplacian Operator
	3 Proposed Approach
		3.1 Proposed Architecture
		3.2 Convolution Kernel Initialization
	4 Experiments
		4.1 Case Study MNIST Handwritten Digits
		4.2 Case Study MNIST American Sign Language
		4.3 Case Study Mexican Sign Language Database
	5 Conclusions
	References
Parameter Optimization of a Convolutional Neural Network Using Particle Swarm Optimization
	1 Introduction
	2 Convolutional Neural Network
		2.1 Input Layer
		2.2 Convolution Layer
		2.3 Non-linearity Layer
		2.4 Pooling Layer
		2.5 Classifier Layer
	3 Particle Swarm Optimization
		3.1 Global Best PSO
		3.2 Local Best PSO
	4 Proposed Method
		4.1 Parameter Optimization of the CNN
		4.2 CNN-PSO Optimization Process
	5 Experiments and Results
		5.1 Exploratory Experiment
		5.2 American Sign Language Alphabet (ASL Alphabet) Experiment
		5.3 American Sign Language MNIST Experiment
		5.4 Analysis and Comparison of Results
	6 Conclusion and Future Work
	References
One-Dimensional Bin Packing Problem: An Experimental Study of Instances Difficulty and Algorithms Performance
	1 Introduction
	2 The Bin Packing Problem
		2.1 Instances
		2.2 Index Description
		2.3 Performance Measures
	3 Algorithms
		3.1 First Fit Decreasing (FFD)
		3.2 Best Fit Decreasing (BFD)
		3.3 Minimum Bin Slack (MBS)
		3.4 GGA-CGT
	4 Results
	5 Experimental Analysis
		5.1 Class BPP.25
		5.2 Class BPP.5
		5.3 Class BPP.75
		5.4 Class BPP1
	6 Conclusions and Future Work
	References
Looking for Emotions in Evolutionary Art
	1 Introduction
	2 In Search of Lost Emotions
		2.1 Humans in the EA Loop
	3 Methodology: Analysis of Emotions in the Era  of Evolutionary Art
		3.1 The Line
		3.2 Simplifying the Problem
		3.3 Evospace-Interactive Module
	4 Results
		4.1 Analyzing Formal Elements
		4.2 Are Emotions Properly Understood?
		4.3 Audience Analysis
		4.4 International Art Competitions
	5 Conclusion
	References
Review of Hybrid Combinations of Metaheuristics for Problem Solving Optimization
	1 Introduction
	2 Review of Hybrid or Combined Metaheuristics
	3 Discussion
	4 Conclusions
	References
GPU Accelerated Membrane Evolutionary Artificial Potential Field for Mobile Robot Path Planning
	1 Introduction
	2 Fundamentals
		2.1 Membrane Computing
		2.2 Evolutionary Computation
		2.3 Artificial Potential Field Method
	3 GPU Accelerated MemEAPF
	4 Results
		4.1 Path Planning Results
		4.2 Performance Results
	5 Conclusions
	References
Optimization of the Internet Shopping Problem with Shipping Costs
	1 Introduction
		1.1 Definition of the Problem
	2 The General Structure of the Memetic Algorithm
		2.1 Selection by Tournament
		2.2 Crossover Operator
		2.3 Mutation Operator
		2.4 Local Search
		2.5 Memetic Algorithm (MAIShOP)
	3 Computational Experiments
	4 Conclusions
	References
Multiobjective Algorithms Performance When Solving CEC09 Test Instances
	1 Introduction
	2 Multiobjective Optimization
	3 CEC09 Test Functions
	4 Multiobjective Optimization Algorithms
	5 Performance Metrics of Multiobjective Optimization
	6 Computational Experiments
	7 Conclusion and Future Work
	References
Analysis of the Efficient Frontier of the Portfolio Selection Problem Instance of the Mexican Capital Market
	1 Introduction
	2 Multiobjective Algorithms in Comparison
	3 CellDE
	4 GDE3
	5 IBEA
	6 MOCell
	7 NSGA-II
	8 NSGA-III
	9 OMOPSO
	10 PAES
	11 SPEA2
	12 Computational Experiments
	13 Conclusions
	References
Multi-objective Portfolio Optimization Problem with Trapezoidal Fuzzy Parameters
	1 Introduction
	2 Elements of Fuzzy Theory
		2.1 Fuzzy Sets
		2.2 Generalized Fuzzy Numbers
		2.3 Addition Operator
		2.4 Graded Mean Integration (GMI)
		2.5 Order Relation in the Set of the Trapezoidal Fuzzy Numbers
		2.6 Pareto Dominance
	3 Multi-objective Portfolio Optimization Problem with Trapezoidal Fuzzy Parameters
	4 Proposal Algorithm T-NSGA-II
		4.1 Representation of the Solutions
		4.2 Evaluating the Solutions
		4.3 One-Point Crossover Operator
		4.4 Uniform Mutation Operator
		4.5 Initial Population
		4.6 Population Sorting
		4.7 No-Dominated Sorting
		4.8 Calculating the Crowding Distance (Deb et al. 2000)
		4.9 Calculating the Spatial Spread Deviation (SSD) (Santiago et al. 2019)
		4.10 Pseudocode of the T-NSGA-II Algorithm
	5 Proposed Strategy to Assess the Performance of Multi-objective Algorithms in the Fuzzy Trapezoidal Numbers Domain
	6 Computational Experiments
	7 Conclusions
	References
A Study on the Use of Hyper-heuristics Based on Meta-Heuristics for Dynamic Optimization
	1 Introduction
	2 Background and Definitions
		2.1 Dynamic Multi-objective Optimization Problem
		2.2 Dynamic Multi-objective Evolutionary Algorithm
		2.3 Hyper-heuristic
		2.4 Indicators to Evaluate DMOEAs Performance Over DMOPs
	3 Relevant Properties to Consider from DMOPs
		3.1 Objective Function
		3.2 Decision Variables
		3.3 Constraints
	4 Known Hyper-heuristic Approaches Towards Solving DOPs
	5 Proposed Checklist and Design Guide for Dynamic Hyper-heuristics
	6 Case Studies Using the Proposed Guide and Checklist
		6.1 Case Study 1
		6.2 Case Study 2
	7 Conclusions and Future Work
	References
On the Adequacy of a Takagi–Sugeno–Kang Protocol as an Empirical Identification Tool for Sigmoidal Allometries in Geometrical Space
	1 Introduction
	2 Methods
		2.1 Model of Complex Allometry
		2.2 TSK Fuzzy Model
		2.3 Data
		2.4 Reproducibility Assessment
		2.5 TSK Identification Procedures
		2.6 Piecewise-Linear Schemes
	3 Results
	4 Discussion
	5 Conclusion
	References
A New Hybrid Method Based on ACO and PSO with Fuzzy Dynamic Parameter Adaptation for Modular Neural Networks Optimization
	1 Introduction
	2 Proposed Method
		2.1 Ant System and ACO Algorithm
		2.2 Particle Swarm Optimization
		2.3 Hybrid Proposed Method
	3 The Neural Network Architectures and Representation
		3.1 Face Database
		3.2 Local Binary Pattern
		3.3 Neural Network Representation for Optimization
		3.4 Neural Network Architectures
	4 Simulation Results
		4.1 Simulation Results for an Artificial Neural Network
		4.2 Simulation Results for a MNN of 2 Modules
		4.3 Simulation Results for a MNN of 3 Modules
		4.4 Simulation Results for a MNN with 4 Modules
		4.5 Statistical Comparison
	5 Conclusions
	References
Knowledge Discovery Using an Evolutionary Algorithm and Compensatory Fuzzy Logic
	1 Introduction
	2 Background and Definitions
		2.1 Knowledge Discovery in Databases
		2.2 Genetic Programming
		2.3 Compensatory Fuzzy Logic
	3 Solution Methodology
		3.1 Knowledge Discovery Algorithm Using Compensatory Fuzzy Logic
		3.2 Generalized Continuous Linguistic Variable Algorithm
	4 Experimentation
	5 Conclusions
	References