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 TinyML and Embedded AI
- Key features of deploying TinyML models
- Limits and requirements of microcontroller-based setups
- Review of essential embedded AI development frameworks
Foundations of Model Optimization
- Analyzing performance bottlenecks in computation
- Pinpointing operations that heavily consume memory
- Establishing baseline performance metrics
Techniques for Quantization
- Strategies for post-training quantization
- Methods for quantization-aware training
- Balancing model accuracy against resource usage
Model Pruning and Compression
- Approaches for both structured and unstructured pruning
- Leveraging weight sharing and model sparsity
- Applying compression algorithms for lightweight inference
Hardware-Specific Optimization
- Deploying models on ARM Cortex-M architectures
- Tailoring optimization for DSP and accelerator units
- Considerations for memory mapping and data flow
Benchmarking and Verification
- Analyzing latency and throughput
- Measuring power draw and energy efficiency
- Testing model accuracy and robustness
Deployment Pipelines and Tooling
- Utilizing TensorFlow Lite Micro for embedded environments
- Incorporating TinyML models into Edge Impulse workflows
- Conducting tests and debugging on physical hardware
Advanced Optimization Methods
- Applying neural architecture search to TinyML
- Combining quantization and pruning strategies
- Using model distillation for embedded inference
Recap and Future Directions
Requirements
- Knowledge of machine learning workflows
- Practical experience with embedded systems or microcontroller development
- Proficiency in Python programming
Intended Audience
- Researchers in artificial intelligence
- Engineers specializing in embedded machine learning
- Professionals developing inference systems with limited resources
