Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Course Outline
Introduction to AI Threat Modeling for Government
- Factors that make AI systems vulnerable in the public sector
- Comparison of AI attack surfaces with traditional systems
- Key attack vectors: data, model, output, and interface layers
Adversarial Attacks on AI Models for Government
- Understanding adversarial examples and perturbation techniques in government applications
- Differentiating between white-box and black-box attacks in public sector systems
- Exploring methods such as FGSM, PGD, and DeepFool in the context of government AI models
- Techniques for visualizing and crafting adversarial samples for enhanced security
Model Inversion and Privacy Leakage for Government
- Methods for inferring training data from model output in government systems
- Membership inference attacks on public sector datasets
- Assessing privacy risks in classification and generative models used by government agencies
Data Poisoning and Backdoor Injections for Government
- Impact of poisoned data on model behavior in government applications
- Addressing trigger-based backdoors and Trojan attacks in public sector AI systems
- Strategies for detection and sanitization to ensure secure operations
Robustness and Defense Techniques for Government
- Implementing adversarial training and data augmentation in government models
- Utilizing gradient masking and input preprocessing techniques in public sector AI
- Applying model smoothing and regularization methods to enhance security
Privacy-Preserving AI Defenses for Government
- Introduction to differential privacy in government applications
- Techniques for noise injection and managing privacy budgets in public sector data
- Leveraging federated learning and secure aggregation methods in government AI systems
AI Security in Practice for Government
- Conducting threat-aware model evaluation and deployment in the public sector
- Utilizing ART (Adversarial Robustness Toolbox) in applied government settings
- Analyzing industry case studies of real-world breaches and mitigations relevant to government operations
Summary and Next Steps for Government
Requirements
- A comprehensive understanding of machine learning workflows and model training for government applications
- Practical experience with Python and widely used ML frameworks such as PyTorch or TensorFlow, tailored to meet the needs of public sector projects
- Familiarity with fundamental security and threat modeling concepts is beneficial for ensuring robust governance and accountability in model development
Audience
- Machine learning engineers working in government agencies
- Cybersecurity analysts focused on protecting public sector systems
- AI researchers and model validation teams dedicated to advancing secure and reliable solutions for government use
14 Hours
