AutoML for Embedded

機能と利点

Automation:

• Automated ML Pipeline: Handles data preprocessing, model architecture search, and hyperparameter tuning automatically
• Rapid Prototyping: Enables quick iterations and comprehensive experimentation to refine AI models
• Neural Architecture Search: Finds optimal model architectures for your specific platform constraints

Integration:

• VS Code Plugin: Powerful development environment that leverages the robust Kenning framework from Antmicro
• CodeFusion Studio™ Integration: Seamless workflow integration for embedded software development
• Supports Renode-based simulation and Zephyr RTOS: Facilitates comprehensive testing and deployment in simulated and real-time environments.

Optimization:

• ADI Hardware Acceleration: Specialized optimizations for MAX78002 (AI8X runtime, CNN accelerator) & MAX32690 (TFLite Micro, microTVM) platforms
• Resource-Aware Models: Generates lightweight models optimized for microcontrollers and edge devices
• Constraint-Aware Training: Ensures models perform efficiently within memory, power, and processing limitations

User Experience:

• Intuitive interface accessible to non-experts: Simplifies AI model development for users without extensive data science knowledge.
• Provides detailed performance metrics and reports: Offers insights into model effectiveness and areas for improvement.