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
Foundations of TinyML and Embedded AI
- Key attributes of TinyML model deployment
- Limitations inherent to microcontroller environments
- Overview of essential embedded AI toolchains
Core Principles of Model Optimization
- Recognizing computational bottlenecks
- Identifying operations that heavily impact memory
- Establishing baseline performance profiles
Quantization Methodologies
- Strategies for post-training quantization
- Implementing quantization-aware training
- Assessing the trade-off between accuracy and resource usage
Pruning and Model Compression
- Techniques for 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
- Tuning for DSP and accelerator enhancements
- Considerations for memory mapping and dataflow
Performance Benchmarking and Validation
- Analyzing latency and throughput metrics
- Measuring power and energy consumption
- Testing for accuracy and system robustness
Deployment Workflows and Tooling
- Leveraging TensorFlow Lite Micro for embedded integration
- Incorporating TinyML models into Edge Impulse workflows
- Testing and troubleshooting on physical hardware
Advanced Optimization Tactics
- Applying neural architecture search to TinyML
- Combining quantization with pruning for hybrid approaches
- Utilizing model distillation for embedded inference
Conclusions and Future Directions
Requirements
- Foundational knowledge of machine learning workflows
- Proficiency in embedded systems or microcontroller-based development
- Strong working knowledge of Python programming
Target Audience
- AI Researchers
- Embedded Machine Learning Engineers
- Professionals specializing in resource-limited inference architectures
