| Titre : | Metaheuristic Algorithms. |
| Auteurs : | Mohamed El Amir Sayah, Auteur ; Fatiha Ghedjemis, Auteur |
| Type de document : | Mémoire magistere |
| Editeur : | Biskra [Algérie] : Faculté des Sciences Exactes et des Sciences de la Nature et de la Vie, Université Mohamed Khider, 2025 |
| Format : | 1VOL.(48p) / ill.couv.ill.en coul / 30cm |
| Langues: | Anglais |
| Mots-clés: | Metaheuristic Algorithms,Genetic Algorithms,Optimization. Adversarial Attacks,Deep Neural Networks |
| Résumé : |
This thesis explores Genetic Algorithms (GAs) for complex optimization, focusing on black-box adversarial attacks against Deep Neural Networks (DNNs). These attacks generate subtle input perturbations to induce misclassification. GAs are well-suited due to their gradient-free nature and ability to navigate complex search spaces. The POBA-GA framework is introduced, customizing GA components like a specialized fitness function (balancing attack success and minimal perturbation) and a novel perceptibility metric, Z(A). POBA-GA's success highlights GAs' versatility, machine learning vulnerabilities, and the need for robust defense mechanisms. |
| Sommaire : |
Dedicace iii Acknowledgements iii Contents iv List of Figures vi Notation viii 1 Optimization problem 1 1.1 What are optimization problems and their importance . . . . . . . . . . . . . . 1 1.1.1 Defining Optimization Problems . . . . . . . . . . . . . . . . . . . . . . . 1 1.1.2 The Importance of Optimization Problems . . . . . . . . . . . . . . . . . 3 1.2 Challenges in optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.2.1 Why metaheuristic algorithms are crucial and how the traditional optimization methods are restricted . . . . . . . . . . . . . . . . . . . . . . . 8 1.3 Metaheuristic algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.3.1 Introduction to metaheuristic algorithms and their significance . . . . . . 10 2 The Genetic Algorithm 12 2.1 Mathematical Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.1.1 Representation (Encoding) . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.1.2 Population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.1.3 Fitness Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.2 Evolutionary Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.2.1 Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.2.2 Crossover (Recombination) . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.2.3 Mutation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.3 Algorithmic Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.4 Theoretical Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.5 Illustrative Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.5.1 Optimization of Test Functions . . . . . . . . . . . . . . . . . . . . . . . 15 2.5.2 Combinatorial Optimization: The Traveling Salesperson Problem (TSP) 17 3 Applying Genetic Algorithms to Adversarial Attacks on Neural Networks 18 3.1 Problem Definition for Adversarial Attacks . . . . . . . . . . . . . . . . . . . . . 18 3.2 The POBA-GA Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.3 Genetic Algorithm Components in POBA-GA . . . . . . . . . . . . . . . . . . . 20 3.3.1 Initialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.3.2 Fitness Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.3.3 Evolutionary Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.3.4 Generation Update and Termination . . . . . . . . . . . . . . . . . . . . 23 3.4 Perturbation Evaluation Metric in POBA-GA . . . . . . . . . . . . . . . . . . . 23 3.5 Demonstration of Successful Attacks . . . . . . . . . . . . . . . . . . . . . . . . 23 Bibliography 27 |
| Type de document : | Mémoire master |
Disponibilité (1)
| Cote | Support | Localisation | Statut |
|---|---|---|---|
| MM/1377 | Mémoire master | bibliothèque sciences exactes | Consultable |
