FEMBA on the Edge: Physiologically-Aware Pre-Training, Quantization, and Deployment of a Bidirectional Mamba EEG Foundation Model on an Ultra-Low Power Microcontroller

Key Points

• The aim is to create a framework for deploying EEG models on low-power devices for health monitoring.
• Developed a full-stack framework for EEG model deployment
• Focused on ultra-low-power microcontroller integration
• Implemented pre-training and quantization techniques
• Enabled continuous monitoring of epilepsy and sleep disorders
• Achieved functionality on wearable technology