Optimization and Machine Learning

Rachid Chelouah has a PhD and a Doctorate of Sciences (Habilitation) from CY Cergy Paris University, France. His main research interests are data science optimization and artificial intelligence methods and their applications in various fields of IT engineering, health, energy and security.Patrick Siarry is a Professor in automatics and informatics at Paris-East Creteil University, France. His main research interests are the design of stochastic global optimization heuristics and their applications in various engineering fields. He has coordinated several books in the field of optimization.

• Introduction xiRachid CHELOUAHPart 1 Optimization 1Chapter 1 Vehicle Routing Problems with Loading Constraints: An Overview of Variants and Solution Methods 3Ines SBAI and Saoussen KRICHEN1.1 Introduction 31.2 The capacitated vehicle routing problem with two-dimensional loading constraints 51.2.1 Solution methods 61.2.2 Problem description 81.2.3 The 2L-CVRP variants 91.2.4 Computational analysis 101.3 The capacitated vehicle routing problem with three-dimensional loading constraints 111.3.1 Solution methods 111.3.2 Problem description 131.3.3 3L-CVRP variants 141.3.4 Computational analysis 161.4 Perspectives on future research 181.5 References 18Chapter 2 MAS-aware Approach for QoS-based IoT Workflow Scheduling in Fog-Cloud Computing 25Marwa MOKNI and Sonia YASSA2.1 Introduction 262.2 Related works 272.3 Problem formulation 292.3.1 IoT-workflow modeling 312.3.2 Resources modeling 312.3.3 QoS-based workflow scheduling modeling 312.4 MAS-GA-based approach for IoT workflow scheduling 332.4.1 Architecture model 332.4.2 Multi-agent system model 342.4.3 MAS-based workflow scheduling process 352.5 GA-based workflow scheduling plan 382.5.1 Solution encoding 392.5.2 Fitness function 412.5.3 Mutation operator 412.6 Experimental study and analysis of the results 432.6.1 Experimental results 452.7 Conclusion 512.8 References 51Chapter 3 Solving Feature Selection Problems Built on Population-based Metaheuristic Algorithms 55Mohamed SASSI3.1 Introduction 563.2 Algorithm inspiration 573.2.1 Wolf pack hierarchy 573.2.2 The four phases of pack hunting 583.3 Mathematical modeling 593.3.1 Pack hierarchy 593.3.2 Four phases of hunt modeling 613.3.3 Research phase – exploration 643.3.4 Attack phase – exploitation 653.3.5 Grey wolf optimization algorithm pseudocode 663.4 Theoretical fundamentals of feature selection 673.4.1 Feature selection definition 673.4.2 Feature selection methods 683.4.3 Filter method 683.4.4 Wrapper method 693.4.5 Binary feature selection movement 693.4.6 Benefits of feature selection for machine learning classification algorithms 703.5 Mathematical modeling of the feature selection optimization problem 703.5.1 Optimization problem definition 713.5.2 Binary discrete search space 713.5.3 Objective functions for the feature selection 723.6 Adaptation of metaheuristics for optimization in a binary search space 763.6.1 Module 𝑀1 773.6.2 Module 𝑀2 783.7 Adaptation of the grey wolf algorithm to feature selection in a binary search space 813.7.1 First algorithm bGWO1 813.7.2 Second algorithm bGWO2 833.7.3 Algorithm 2: first approach of the binary GWO 843.7.4 Algorithm 3: second approach of the binary GWO 853.8 Experimental implementation of bGWO1 and bGWO2 and discussion 863.9 Conclusion 873.10 References 88Chapter 4 Solving the Mixed-model Assembly Line Balancing Problem by using a Hybrid Reactive Greedy Randomized Adaptive Search Procedure 91Belkharroubi LAKHDAR and Khadidja YAHYAOUI4.1 Introduction 924.2 Related works from the literature 954.3 Problem description and mathematical formulation 974.3.1 Problem description 974.3.2 Mathematical formulation 984.4 Basic greedy randomized adaptive search procedure 994.5 Reactive greedy randomized adaptive search procedure 1004.6 Hybrid reactive greedy randomized adaptive search procedure for the mixed model assembly line balancing problem type-2 1014.6.1 The proposed construction phase 1024.6.2 The local search phase 1064.7 Experimental examples 1074.7.1 Results and discussion 1114.8 Conclusion 1154.9 References 116Part 2 Machine Learning 119Chapter 5 An Interactive Attention Network with Stacked Ensemble Machine Learning Models for Recommendations 121Ahlem DRIF, SaadEddine SELMANI and Hocine CHERIFI5.1 Introduction 1225.2 Related work 1245.2.1 Attention network mechanism in recommender systems 1245.2.2 Stacked machine learning for optimization 1255.3 Interactive personalized recommender 1265.3.1 Notation 1285.3.2 The interactive attention network recommender 1295.3.3 The stacked content-based filtering recommender 1345.4 Experimental settings 1365.4.1 The datasets 1365.4.2 Evaluation metrics 1375.4.3 Baselines 1395.5 Experiments and discussion 1405.5.1 Hyperparameter analysis 1405.5.2 Performance comparison with the baselines 1435.6 Conclusion 1465.7 References 146Chapter 6 A Comparison of Machine Learning and Deep Learning Models with Advanced Word Embeddings: The Case of Internal Audit Reports 151Gustavo FLEURY SOARES and Induraj PUDHUPATTU RAMAMURTHY6.1 Introduction 1526.2 Related work 1546.2.1 Word embedding 1566.2.2 Deep learning models 1576.3 Experiments and evaluation 1586.4 Conclusion and future work 1636.5 References 165Chapter 7 Hybrid Approach based on Multi-agent System and Fuzzy Logic for Mobile Robot Autonomous Navigation 169Khadidja YAHYAOUI7.1 Introduction 1707.2 Related works 1717.2.1 Classical approaches 1727.2.2 Advanced methods 1737.3 Problem position 1747.4 Developed control architecture 1767.4.1 Agents description 1777.5 Navigation principle by fuzzy logic 1837.5.1 Fuzzy logic overview 1837.5.2 Description of simulated robot 1847.5.3 Strategy of navigation 1857.5.4 Fuzzy controller agent 1867.6 Simulation and results 1947.7 Conclusion 1967.8 References 196Chapter 8 Intrusion Detection with Neural Networks: A Tutorial 201Alvise DE’ FAVERI TRON8.1 Introduction 2018.1.1 Intrusion detection systems 2018.1.2 Artificial neural networks 2028.1.3 The NSL-KDD dataset 2028.2 Dataset analysis 2038.2.1 Dataset summary 2038.2.2 Features 2038.2.3 Binary feature distribution 2048.2.4 Categorical features distribution 2078.2.5 Numerical data distribution 2118.2.6 Correlation matrix 2128.3 Data preparation 2138.3.1 Data cleaning 2138.3.2 Categorical columns encoding 2138.3.3 Normalization 2148.4 Feature selection 2178.4.1 Tree-based selection 2178.4.2 Univariate selection 2188.5 Model design 2198.5.1 Project environment 2198.5.2 Building the neural network 2208.5.3 Learning hyperparameters 2208.5.4 Epochs 2208.5.5 Batch size 2218.5.6 Dropout layers 2218.5.7 Activation functions 2228.6 Results comparison 2228.6.1 Evaluation metrics 2228.6.2 Preliminary models 2238.6.3 Adding dropout 2258.6.4 Adding more layers 2268.6.5 Adding feature selection 2278.7 Deployment in a network 2288.7.1 Sensors 2288.7.2 Model choice 2298.7.3 Model deployment 2298.7.4 Model adaptation 2318.8 Future work 2318.9 References 231List of Authors 233Index 235