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 Pipelines
- Explore the distinct stages involved in the TinyML workflow
- Examine the defining characteristics of edge hardware
- Consider key aspects of pipeline design
Data Collection and Preprocessing
- Gather structured data and sensor inputs
- Implement data labeling and augmentation techniques
- Prepare datasets tailored for constrained environments
Model Development for TinyML
- Choose appropriate model architectures for microcontrollers
- Execute training workflows with standard ML frameworks
- Assess various model performance indicators
Model Optimization and Compression
- Apply quantization methods
- Utilize pruning and weight sharing strategies
- Balance model accuracy against resource limitations
Model Conversion and Packaging
- Export models to TensorFlow Lite
- Incorporate models into embedded toolchains
- Handle model size and memory constraints effectively
Deployment on Microcontrollers
- Flash models onto specific hardware targets
- Set up the necessary run-time environments
- Conduct real-time inference tests
Monitoring, Testing, and Validation
- Develop testing strategies for deployed TinyML systems
- Debug model behavior directly on hardware
- Verify performance under actual field conditions
Integrating the Full End-to-End Pipeline
- Create automated workflows
- Manage versioning for data, models, and firmware
- Oversee updates and iterative improvements
Summary and Next Steps
Requirements
- A solid grasp of machine learning fundamentals
- Practical experience with embedded programming
- Knowledge of Python-based data workflows
Audience
- AI engineers
- Software developers
- Embedded systems experts
