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.
Course Outline
Overview of Edge AI and Model Optimization
- Grasping the nature of edge computing and associated AI workloads
- Balancing performance metrics against resource limitations
- Surveying various model optimization approaches
Model Selection and Pre-training
- Evaluating lightweight architectures (e.g., MobileNet, TinyML, SqueezeNet)
- Analyzing model structures ideal for edge devices
- Leveraging pre-trained models as a foundational base
Fine-Tuning and Transfer Learning
- Core principles of transfer learning
- Adapting models to proprietary datasets
- Executing practical fine-tuning processes
Model Quantization
- Methods for post-training quantization
- Quantization-aware training techniques
- Assessing accuracy trade-offs and performance
Model Pruning and Compression
- Differentiating between structured and unstructured pruning strategies
- Techniques for compression and weight sharing
- Conducting benchmarks on compressed models
Deployment Frameworks and Tools
- Exploring TensorFlow Lite, PyTorch Mobile, and ONNX
- Ensuring compatibility with edge hardware and runtime settings
- Utilizing toolchains for cross-platform deployment
Practical Deployment
- Deploying models to Raspberry Pi, Jetson Nano, and mobile devices
- Performing profiling and benchmarking tasks
- Resolving common deployment challenges
Recap and Future Directions
Requirements
- A solid grasp of fundamental machine learning concepts
- Proficiency in Python and popular deep learning frameworks
- Awareness of embedded systems and the limitations of edge devices
Target Audience
- Developers specializing in embedded AI
- Experts in edge computing architectures
- Machine learning engineers with a focus on edge-based deployment
14 Hours