Course Details
Course Code: CDAIP ;Â Â Duration: 5 Days; Instructor-led
Outcomes
Upon successful completion of this course, students will be able to:
- Understand the different attacks on Large Language Models (LLMs), Deep Learning Models (DLMs), and on Tree-Ensemble Models and Forecasting.
- Understand the different types of AI exploitation techniques, including model inversion attacks, adversarial examples, data poisoning, and model extraction.
- Analyze the risks and vulnerabilities associated with AI systems and develop strategies to mitigate them.
- Design and implement effective defense mechanisms to protect AI modules from attacks by other AI modules.
Audience
- Data Science Analysts / Profesisonals
- AI Engineers
- AI Developers (LLM, GenAI, etc)
- AI Architects
- AI designer
- AI ethics specialists
- Pentesters
- Security Analysts
- Bug Bountry Hunters
- Security Consultants
- Blue Team members, Defenders,
- and Forensic Analyst
Prerequisites
Students enrolling in this course should have a basic understanding of cybersecurity principles and AI concepts. Familiarity with machine learning algorithms and Python will be beneficial but not mandatory.
Methodology
This program will be conducted with interactive lectures, PowerPoint presentation, discussion and practical exercise.
Course Objectives
The objective of this course is to empower professionals with the requisite knowledge and skills to safeguard AI modules from attacks by Cybercriminals or other AI modules. This course delves deep into the complex world of AI-driven threats, providing a comprehensive understanding of the techniques and strategies used to counteract them.
The primary objective of this course is to equip learners with a deep understanding of Adversarial AI and its different techniques and how they can leverage AI to protect AI models. This includes learning how to:
- Understand the concepts and techniques used to exploit AI modules, including adversarial attacks, data poisoning, and model inversion attacks, with permission from the system owners. Identify potential vulnerabilities in AI-powered systems and develop strategies to prevent exploitation by malicious actors.
- Implement effective defence mechanisms to protect AI modules from attacks launched by other AI systems.
- Develop a comprehensive understanding of the AI security landscape, including the latest threats, trends, and best practices in AI defence.
Outlines
- Overview of AI and Machine Learning Concepts
- Types of AI Models and Architectures
- AI Development Lifecycle and Workflows
- AI Ethics and Responsible AI Principles
- Overview of AI Security Landscape
- Common Attack Vectors on AI Models
- Threat Modelling for AI Systems
- AI Security Best Practices and Frameworks
- Attacks on Large Language Models (LLMs)
- Attacks on Deep Learning Models
- Attacks on Tree-Ensemble Models and Forecasting
- Data Poisoning and Manipulation Attacks
- Reconnaissance and Vulnerability Scanning
- Exploiting Vulnerabilities in AI Infrastructure
- Attacks on AI APIs and Interfaces
- Jailbreaking LLMs and Diffusion Models
- Membership Inference Attacks
- Model Inversion and Extraction Attacks
- Adversarial Defenses and Robustness
- Course Recap
- Assessment
