Ultra-Small Footprint, Ultra-Low Power and Ultra-Efficient ML Models for Extreme Edge IoT Applications
Microchip and SensiML have partnered to bring a powerful but easy to use integration of SensiML industry-leading AutoML toolkit for smart sensing to Microchip’s full range of MCUs. The integration leverages Microchip’s world-class MPLAB IDE along with the power and efficiency only SensiML can offer for creating extremely compact models down to 8-bit platforms.
Microchip’s ML plugin for Data Visualizer, a powerful debug tool, allows users to import data streams directly into SensiML Data Capture Lab making the process of collecting data into SensiML simple and seamless with MPLAB.
On the outbound side, SensiML Analytics Studio can generate autonomous machine learning models called Knowledge Packs that provide autonomous insight directly on the IoT microcontroller. SensiML Knowledge Packs can be output in binary, library, or C source code form. The binary option is useful for quick testing of the ML model itself while source code provides maximum flexibility. Thus a virtuous cycle of capture, label, code-gen, test can be repeated rapidly to yield highly accurate and compact smart sensing models with a minimum of time and ramp-up.
- Supports 32-bit PIC and SAM microcontrollers as well as 16-bit PIC24 and dsPIC, and 8-bit AVR processors
- Small memory footprint combined with a high degree of accuracy
- Makes it easy to upgrade legacy applications with AI capabilities
- Extremely low power consumption
- Edge AI IoT without the need for cloud connectivity
The ability of the SensiML Analytics Toolkit to easily generate high-quality, power efficient models with a small memory footprint is an excellent fit for our customers that wants to add machine learning to their existing designs.
Supporting products from these leading companies and SensiML partners