Microchip makes it easier to add ML to processors

Microchip Technology has introduced a new software toolkit that can be used across the company’s portfolio of microcontrollers (MCUs) and microprocessors (MPUs), making it easier to add machine-learning (ML) inferences. The new MPLAB Machine Learning Development Suite is a complete integrated workflow for streamlined ML model development for embedded engineers with or without ML programming knowledge.

Microchip said it is the first integrated solution that supports 8-, 16- and 32-bit MCUs and MPUs for ML at the edge. The toolkit can be used for a range of applications including industrial, manufacturing, consumer and automotive.

(Source: Microchip Technology)

Developed to help embedded designers build small-footprint ML models, the new development suite, powered by AutoML, eliminates many repetitive and time-consuming model-building tasks, including extraction, training, validation and testing, said Microchip. It also provides model optimizations that consider the memory constraints of MCU and MPUs.

When used in combination with the  MPLAB X Integrated Development Environment (IDE), the new toolkit provides a complete solution that can be easily implemented by designers with little to no ML programming knowledge, Microchip said, and it is also sophisticated enough for more experienced ML designers.

Designers can optionally bring a model from TensorFlow Lite and use it in any MPLAB Harmony v3 project. The embedded software development framework provides interoperable software modules to simplify development.

Microchip also offers the VectorBlox Accelerator Software Development Kit (SDK) for power-efficient convolutional neural network (CNN)-based artificial intelligence/machine learning inference with PolarFire FPGAs.

Pricing for the new development suite varies based on licensing. A free version is available for evaluation. Visit www.microchipdirect.com for purchasing information.