New Perspectives on Hybrid Intelligent System Design based on Fuzzy Logic, Neural Networks and Metaheuristics

Preface
About This Book
Contents
Neural Networks
Automated Medical Diagnosis and Classification of Skin Diseases Using Efficinetnet-B0 Convolutional Neural Network
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 EfficientNet-B0
		3.2 ResNet50
		3.3 CNN
		3.4 Dataset
		3.5 Experimental Setup
	4 Results and Discussion
	5 Limitation of the Study
	6 Conclusion and Future Scope
	References
Modular Approach for Neural Networks in Medical Image Classification with Enhanced Fuzzy Integration
	1 Introduction
		1.1 Medical Images
		1.2 Medical Image Classification Using Neural Networks
		1.3 Fuzzy Logic Combined with Modular Neural Networks
	2 Proposed Method
	3 Methodology
	4 Results and Discussion
	5 Conclusions
	References
Clustering and Prediction of Time Series for Traffic Accidents Using a Nested Layered Artificial Neural Network Model
	1 Introduction
	2 Case Study
	3 Methodology
	4 Experiments and Results
	5 Conclusions
	References
Ensemble Recurrent Neural Networks and Their Optimization by Particle Swarm for Complex Time Series Prediction
	1 Introduction
	2 Problem Statement and Proposed Method
		2.1 Description of the Particle Swarm Optimization Applied to Recurrent Neural Network
		2.2 Data Base
		2.3 Description of Type-1 and IT2FS
	3 Simulation Results
	4 Conclusions
	References
Filter Estimation in a Convolutional Neural Network with Type‐2 Fuzzy Systems and a Fuzzy Gravitational Search Algorithm
	1 Introduction
	2 Literature Review
		2.1 Convolutional Neural Networks
		2.2 Type-2 Fuzzy Logic System
	3 Proposed Method
	4 Results and Discussion
	5 Conclusions
	References
Optimization
Artificial Fish Swarm Algorithm for the Optimization of a Benchmark Set of Functions
	1 Introduction
	2 Related Works
	3 Artificial Fish Swarm Algorithm
	4 Benchmark Sets of Functions
	5 Experimental Results
	6 Analysis of the Parameters in AFSA
	7 Comparison with Others Metaheuristics
	8 Conclusions
	References
Hierarchical Logistics Methodology for the Routing Planning of the Package Delivery Problem
	1 Introduction
	2 Problem Definition
	3 Hierarchical Logistics Methodology
		3.1 Phase 1: Clustering by FCM
		3.2 Phase2: Optimization by ACO
	4 Computational Experiment
	5 Conclusions
	References
A Novel Distributed Nature-Inspired Algorithm for Solving Optimization Problems
	1 Introduction
	2 Proposal
		2.1 Birth
		2.2 Growth
		2.3 Reproduction
		2.4 Death
	3 Experiments
		3.1 Experimental Setup
		3.2 Experiment Configuration
		3.3 Experiment Results
	4 Discussion
	5 Conclusions
	References
Evaluation and Comparison of Brute-Force Search and Constrained Optimization Algorithms to Solve the N-Queens Problem
	1 Introduction
	2 Related Work
	3 Mathematical Models
		3.1 Mathematical Model No. 1
		3.2 Mathematical Model No. 2
	4 N-Queens Problem Solution Algorithms
		4.1 Backtracking Algorithm
		4.2 Branch and Bound Algorithm
		4.3 Linear Programming Algorithm
	5 Experiments and Results
		5.1 Experiment 1: Backtracking
		5.2 Experiment 2: Branch and Bound
		5.3 Experiment 3: Linear Programming
	6 Conclusions and Future Work
	References
Performance Comparative Between Single and Multi-objective Algorithms for the Capacitated Vehicle Routing Problem
	1 Introduction
	2 Important Concepts
		2.1 Continuous Optimization
		2.2 Multi-objective Optimization
		2.3 Multi-objective Genetic Algorithms
		2.4 Nsga-Ii
		2.5 Wasfga
		2.6 Library
	3 Mono-objective Approach
		3.1 Class’s Description
		3.2 Restrictions
	4 Methodology
	5 Experiments
	6 Results
	7 Multi-objective Approach
		7.1 Fitness Functions
		7.2 Methodology
		7.3 Results
	8 Conclusions
	References
Fuzzy Logic
Optimization of a Fuzzy Classifier for Obtaining the Blood Pressure Levels Using the Ant Lion Optimizer
	1 Introduction
	2 Literature Review
		2.1 Ant Lion Optimizer
		2.2 Type-1 Fuzzy System
		2.3 Blood Pressure and Hypertension
	3 Proposed Method
	4 Results
	5 Conclusions
	References
Optimization of Fuzzy-Control Parameters for Path Tracking of a Mobile Robot Using Distributed Genetic Algorithms
	1 Introduction
	2 Proposed Method
		2.1 Distributed GA
		2.2 Membership Function Optimization
		2.3 Four Parameters Configuration
		2.4 Nine Parameters Configuration
		2.5 18 Parameter Configuration
		2.6 Implementation
	3 Experiments and Results
		3.1 Fuzzy Control
		3.2 GA Setup
	4 Conclusions
	References
A New Fuzzy Approach to Dynamic Adaptation of the Marine Predator Algorithm Parameters in the Optimization of Fuzzy Controllers for Autonomous Mobile Robots
	1 Introduction
	2 Fuzzy Logic and Marine Predators Algorithm
		2.1 Marine Predator Algorithm
		2.2 MPA Formulation
		2.3 Fuzzy Marine Predator Algorithm (FMPA)
	3 Fuzzy Logic Systems
	4 Study Cases
		4.1 Case 1: Benchmark CEC-2017 Functions
		4.2 Case 2: Optimization of Fuzzy Controllers
	5 Results
		5.1 Case 1 Results: Benchmark CEC-2017 Functions
		5.2 Case 2: Dynamic Adjustment of Fuzzy Controller Parameters
		5.3 Statistical Comparison
	6 Conclusions
	References
Evaluation of Times and Best Solutions of MFO, LSA and PSO Using Parallel Computing, Fuzzy Logic Systems and Migration Blocks Together to Evaluate Benchmark Functions
	1 Introduction
	2 Particle Swarm Optimization (PSO)
	3 Moth Flame Optimization (MFO)
	4 Lightning Search Algorithm (LSA)
	5 Parallel MFO, LSA and PSO Algorithms with Fuzzy Systems and Migration
	6 Fuzzy Logic System
	7 Experiments
		7.1 Experimental Results with PSO
	8 Conclusions
	References
Fuzzy Dynamic Parameter Adaptation in the Mayfly Algorithm: Preliminary Tests for a Parameter Variation Study
	1 Introduction
	2 Theoretical Framework
	3 Proposed Method
	4 Parameter Impact Study
	5 Results
	6 Conclusions
	References
Optimization: Theory and Applications
Symmetric-Approximation Energy-Based Estimation of Distribution (SEED) Algorithm for Solving Continuous High-Dimensional Global Optimization Problems
	1 Introduction
	2 Background
		2.1 Continuous Global Optimization Problems.
		2.2 Symmetric-Approximation Energy-Based Estimation of Distribution (SEED)
		2.3 Differential Evolution
		2.4 Particle Swarm Optimization
	3 Experimental Design
	4 Results
		4.1 Page’s Trend Test Statistical Analysis
	5 Conclusions
	References
Optimization Models and Methods for Bin Packing Problems: A Case Study on Solving 1D-BPP
	1 Introduction
	2 The Bin Packing Problems
		2.1 The One-Dimensional Bin Packing Problem and Variants
		2.2 Multiobjective Bin Packing Problem
		2.3 Dynamic Bin Packing
	3 Related Works
	4 Case Study: State-of-the-Art Solution Methods for 1D-BPP
		4.1 Grouping Genetic Algorithm with Controlled Gene Transmission
		4.2 Consistent Neighborhood Search for One-Dimensional Bin Packing
	5 Design of GGA-CGT-II as an Extension of GGA-CGT
		5.1 Design of Strategies and Analysis of Their Impact on Performance
		5.2 Method to Calculate Different Lower Bounds
		5.3 Instance Reduction Method to Simplify the Problem
	6 Experimentations and Results with GGA-CGT-II
		6.1 Instances for 1D-Bin Packing
		6.2 Experiment 1: Calculation of Different Lower Bounds
		6.3 Experiment 2: The Proposed Methods Together (GGA-CGT-II)
	7 Conclusions and Future Work
	References
CMA Evolution Strategy Applied to Optimize Chemical Molecular Clusters MxNz (x + y ≤ 5; M = N or M ≤ N)
	1 Introduction
	2 Potential Energy Surface
	3 Covariance Matrix Adaptation—Evolution Strategy
	4 Experiments
	5 Results
	6 Conclusions and Future Work
	References
Specialized Crossover Operator for the Differential Evolution Algorithm Applied to a Car Sequencing Problem with Constraint Smoothing
	1 Introduction
	2 Differential Evolution with Specialized Chromosome Repairer Algorithm (DECR-s)
	3 Specialized Crossover Operator
		3.1 Specialized Crossover Operator for Chromosome a (Classes)
		3.2 Specialized Crossover Operator for Chromosome B (Color)
	4 Experiment Design
	5 Results
		5.1 Results Per Case of the Crossover Operator
		5.2 Statistical Analysis of Cases
	6 Conclusion
	References
A Brave New Algorithm to Maintain the Exploration/Exploitation Balance
	1 Introduction
	2 State of the Art
	3 Algorithm's Nature
	4 Implementation
	5 Experimental Results
		5.1 Diversity Analysis
		5.2 Diversity in Brave New Algorithm
		5.3 Diversity on A Basic Genetic Algorithm
	6 Discussion and Conclusions
	References
A New Optimization Method Based on the Lotka-Volterra System Equations
	1 Introduction
	2 Optimization
	3 Lotka-Volterra System Equations
	4 Proposed Method
	5 Results
	6 Conclusions
	References
Hybrid Intelligent Systems
A Comparison of Replacement Operators in Heuristics for CSP Problems
	1 Introduction
	2 Background
		2.1 Constraint Satisfaction Problems
		2.2 Heuristics
		2.3 Acceptance Criteria
	3 Methodology
	4 Results
		4.1 Graph Coloring
		4.2 Capacitated Vehicle Routing Problem
	5 Discussion and Future Work
	References
Synchronisms Using Reinforcement Learning as an Heuristic
	1 Introduction
	2 Background
		2.1 Synchronisms
		2.2 Agents
		2.3 Machine Learning
		2.4 Deep Learning
		2.5 Reinforcement Learning
		2.6 Q-Learning
	3 Related Work
		3.1 Most Common Methods of Deep Reinforcement Learning
		3.2 Real Life Applications of Deep Reinforcement Learning
	4 Using Synchronisms as an Heuristic for RL
		4.1 RL Model Using Gradient Descent
		4.2 Agent Learning Model
		4.3 Keen Eye
	5 Study Case: Cartpole
		5.1 Defining the Environment
		5.2 Defining the Agents
	6 Results and Discussions
		6.1 Comparing Both Models Using Reliability Metrics
	7 Conclusions and Future Work
		7.1 Future Work
	References
A Mathematical Deduction of Variational Minimum Distance in Gaussian Space and Its Possible Application to Artificial Intelligence
	1 Introduction
	2 Methodology
		2.1 Fisher Information Metric
		2.2 Gaussian Arc Length
		2.3 Variational Minimum Distance in Gaussian Space
	3 Connection Between UMDAc and the Minimum Variational Distance
		3.1 Univariate Marginal Distribution Algoritm
		3.2 Minimum Distance in Gaussian UMDAc
	4 Experimental Setup
		4.1 Fitness Functions and Parameter Configuration
		4.2 Experimental Results
		4.3 Resulting Mathematical Model
	5 Conclusions
	References
A Model for Learning Cause-Effect Relationships in Bayesian Networks
	1 Introduction
	2 Rescorla-Wagner Model: Cognitive Psychology
	3 Causal Bayesian Networks: Artificial Intelligent
	4 The Algorithm CBN-RW
		4.1 Information Measures from Hypothesis Tests
		4.2 Package ‘Ndl’
		4.3 Description of the CBN-RW Algorithm
	5 Selection of Experimental Datasets
	6 Results
	7 Conclusions
	References
Eureka-Universe: A Business Analytics and Business Intelligence System
	1 Introduction
	2 Background
		2.1 Knowledge Discovery
		2.2 Fuzzy Logic
		2.3 Compensatory Fuzzy Logic
		2.4 Archimedean Compensatory Fuzzy Logic
		2.5 Genetic Algorithms
		2.6 Fuzzy Inference
		2.7 Fuzzy Interpretability
	3 Eureka-Universe
		3.1 Architecture
		3.2 Scientific Core
		3.3 Project Manager
		3.4 Task Manager
	4 Case of Study
	5 Conclusions
	References
Neural Networks and Learning
Extension of Windowing as a Learning Technique in Artificial Noisy Domains
	1 Introduction
	2 Noise Modeling
	3 Case Study
		3.1 Data Generation
		3.2 Inductive Algorithms
		3.3 Methodology
	4 Results
	5 Conclusions
	References
Why Rectified Linear Activation Functions? Why Max-Pooling? A Possible Explanation
	1 Formulation of the Problem: An Explanation is Needed
	2 Why Rectified Linear Neurons: Our Explanation
	3 Why Max-pooling
	4 Which Fuzzy Operations?
	References
Localized Learning: A Possible Alternative to Current Deep Learning Techniques
	1 Formulation of the Problem
	2 Why Neural Networks: A Theoretical Explanation
	3 Need to Go Beyond Traditional Neural Networks and Deep Learning
	4 Beyond Deep Learning, Towards Localization
	References
What is a Reasonable Way to Make Predictions?
	1 Formulation of the Problem
		1.1 Making Predictions is Important
		1.2 At First Glance, the Answer to These Questions is Straightforward
		1.3 Situation is Not so Simple
		1.4 This Should be Decided by an Experiment
	2 Analysis of the Problem on a Simplified Case
		2.1 Simplified Case: A Description
		2.2 Case Study
		2.3 Surprising Conclusion
		2.4 What We Discuss in this Paper
	3 General Case
		3.1 Let us Describe the Situation in Precise Terms
		3.2 Prediction Rule Must be Fair
		3.3 Meta-Analysis: Using Prediction Rule to Select Prediction Rule
		3.4 Induction Versus Anti-induction Revisited
		3.5 General Result
	4 Rules Must be Falsifiable
		4.1 An Example Where a Reasonable Prediction Rule is Inconsistent
		4.2 A Problem with Simple Induction
	5 Conclusions and Future Work
		5.1 Predictions: Naive Idea
		5.2 What We Show: Situation is More Complex That it May Appear
		5.3 Future Work
	References