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.
Duration 21 hours
Course Outline
Introduction to Security in TinyML
- Challenges to security in resource-limited ML systems
- Developing threat models for TinyML implementations
- Categorizing risks in embedded AI applications
Data Privacy in Edge AI
- Privacy implications of on-device data processing
- Strategies to reduce data exposure and transfer
- Methods for decentralized data management
Adversarial Attacks on TinyML Models
- Threats involving model evasion and data poisoning
- Manipulating inputs on embedded sensors
- Evaluating vulnerabilities in constrained environments
Hardening Embedded ML Security
- Protection layers for firmware and hardware
- Access control and secure boot protocols
- Best practices for protecting inference pipelines
Privacy-Preserving Techniques for TinyML
- Considerations for quantization and model design regarding privacy
- Methods for on-device anonymization
- Lightweight encryption and secure computation approaches
Secure Deployment and Maintenance
- Secure provisioning procedures for TinyML devices
- Strategies for OTA updates and patch management
- Edge-level monitoring and incident response
Testing and Validating Secure TinyML Systems
- Frameworks for security and privacy testing
- Simulation of real-world attack scenarios
- Considerations for validation and compliance
Case Studies and Practical Applications
- Analyzing security breaches in edge AI ecosystems
- Designing resilient TinyML architectures
- Assessing the balance between performance and protection
Conclusion and Future Directions
Requirements
- Familiarity with embedded system architectures
- Hands-on experience with machine learning workflows
- Foundation in cybersecurity principles
Intended Audience
- Security analysts
- AI developers
- Embedded engineers
Testimonials (1)
The exercises
