Machine Learning for Cyber Agents: Attack and Defence (Advanced Sciences and Technologies for Security Applications)

✍ Scribed by Stanislav Abaimov, Maurizio Martellini

• Publisher: Springer
• Year: 2022
• Tongue: English
• Leaves: 235
• Edition: 1st ed. 2022
• Category: Library

✦ Synopsis

The cyber world has been both enhanced and endangered by AI. On the one hand, the performance of many existing security services has been improved, and new tools created. On the other, it entails new cyber threats both through evolved attacking capacities and through its own imperfections and vulnerabilities. Moreover, quantum computers are further pushing the boundaries of what is possible, by making machine learning cyber agents faster and smarter.

With the abundance of often-confusing information and lack of trust in the diverse applications of AI-based technologies, it is essential to have a book that can explain, from a cyber security standpoint, why and at what stage the emerging, powerful technology of machine learning can and should be mistrusted, and how to benefit from it while avoiding potentially disastrous consequences. In addition, this book sheds light on another highly sensitive area – the application of machine learning for offensive purposes, an aspect that is widely misunderstood, under-represented in the academic literature and requires immediate expert attention.

✦ Table of Contents

Contents
Abbreviations
Disclosure Statement
List of Figures
List of Tables
1 Introduction
1.1 Motivation
1.2 Aim
1.3 Structure
Reference
2 Understanding Machine Learning
2.1 Setting the Scene
2.2 Conceptual and Operational Landscape
2.2.1 Machine Learning as a Concept
2.2.2 Algorithms and Their Application
2.2.3 Models
2.2.4 Methods
2.3 Explainability of Machine Learning
2.3.1 Data Collection
2.3.2 Pre-processing
2.3.3 Training
2.3.4 Prediction
2.3.5 Evaluation and Metrics
2.3.6 Fine-Tuning
2.4 Quantum Machine Learning
2.4.1 Quantum Computers
2.4.2 Main Notions
2.4.3 Specificity of Quantum Machine Learning
2.5 Machine Learning Limitations
2.6 Conclusion
References
3 Defence
3.1 Machine Learning for Cyber Security
3.2 IDS Supporting Human Operators
3.3 Network Security
3.3.1 Packet Parsing-Based Detection
3.3.2 Payload Analysis-Based Detection
3.4 Computer Security
3.4.1 Hardware Behaviour
3.4.2 Operating System
3.4.3 Connected Devices
3.4.4 Software Analysis
3.5 AI-Specific Security Issues
3.5.1 Adversarial Attacks on Artificial Intelligence
3.5.2 Defence Methods Against Adversarial Attacks
3.5.3 Development of Safe Artificial Intelligence Systems
3.5.4 Hybrid Defence
3.6 Conclusion
4 Attack
4.1 Machine Learning for Malware
4.2 Machine Learning Enhancing Cyber Attacks
4.2.1 Phishing
4.2.2 Exploitation
4.2.3 Network Traffic Masquerading
4.2.4 Bots and Botnets
4.2.5 Password Guessing
4.2.6 Ransomware
4.2.7 Cryptomining Malware
4.2.8 Recovery
4.2.9 Cryptanalysis
4.2.10 Forensics Investigation
4.2.11 Attacks Against Hardware
4.3 Weaponizing AI
4.3.1 Machine Learning for Weapons Autonomy
4.3.2 AWS Vulnerabilities
4.4 Conclusion
Reference
5 International Resonance
5.1 Debates Over AI Integration and Governance
5.1.1 Debates Over Technical Issues
5.1.2 Debates Over Legal and Ethical Issues
5.1.3 Debates Over Governance
5.1.4 Debates Over Military Use of AI Offensive Capabilities
5.2 Multilateral Collaboration for Peaceful AI
5.2.1 Europe Fit for Digital Age
5.2.2 African Digital Transformation
5.2.3 ASEAN Digital Masterplan
5.2.4 United Nations Global Agenda for AI
5.3 Conclusion
References
6 Prospects
6.1 Technological Development
6.2 Societal Transformation
7 Conclusion
Glossary
References