3.60. AI362 AI for edge computing (Elective)

3.60. AI362 AI for edge computing (Elective)

Figure 3.59: Connection Map. AI362 AI for edge computing

3.60.1. Justification Back to top

This course covers AI model deployment on edge devices (IoT, mobile, embedded), addressing optimization techniques, specialized frameworks, and privacy/security considerations in distributed systems.

3.60.2. Generales Goals Back to top

  1. Implement AI models optimized for edge hardware
  2. Design distributed inference pipelines
  3. Evaluate accuracy/latency/energy trade-offs

3.60.3. Contribution to Outcomes Back to top

AG-C02) Ethics: Applies ethical principles and commits to professional ethics and the standards of professional computing practice. (Usage)
AG-C09) Design and Development of Solutions: Designs, implements, and evaluates solutions for complex computing problems. (Assessment)
AG-C12) Applies computer science theory and software development fundamentals to produce computer-based solutions. (Assessment)

3.60.4. Content Back to top

3.60.4.1. Edge AI Fundamentals (12 hours) [Skills AG-C09] Back to top

Bibliography: (Han and Lin, 2022)

Topics

  1. Hardware architectures for edge computing
  2. Edge-friendly model taxonomies (MobileNets, TinyML)
  3. Deployment frameworks (TF Lite, ONNX Runtime)

Learning Outcomes

  1. Select models matching hardware constraints [Assessment]
  2. Profile energy consumption on real devices [Usage]
3.60.4.2. Optimization Techniques (24 hours) [Skills AG-C09] Back to top

Bibliography: (Howard et al., 2017; Saraph and Warden, 2021)

Topics

  1. Quantization and pruning
  2. Knowledge distillation
  3. Specialized compilation (TVM, Apache MXNet)

Learning Outcomes

  1. Shrink models for microcontrollers [Assessment]
  2. Implement hybrid (edge+cloud) pipelines [Usage]
3.60.4.3. Security & Privacy (8 hours) [Skills AG-C09] Back to top

Bibliography: (Han and Lin, 2022)

Topics

  1. Federated learning at edge
  2. Homomorphic encryption
  3. Edge model attacks (adversarial examples)

Learning Outcomes

  1. Analyze edge system vulnerabilities [Familiarity]
  2. Implement basic defenses [Usage]

3.60.5. Bibliography Back to top

Han, S. and Lin, Y. (2022). Edge AI: Systems Design for Embedded Machine Learning. Springer.

Howard, A., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., and Adam, H. (2017). Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv:1704.04861.

Saraph, V. and Warden, P. (2021). Tinyml: Enabling always-on ml on the edge. In ACM HotNets.

Spotted a typo, an outdated course, a broken link, or have a suggestion? Let us know.

Scan to open on your phone