Founded in 1935 originally to commercialize ferrite (i.e. a magnetic material invented at the Tokyo Institute of Technology), TDK Global was built on the motto of “Contribute to culture and industry through creativity.” Determined to create unprecedented value and to be of service, TDK has grown and now operates in more than 250 locations in 30 countries and regions, with approximately 117,000 employees worldwide.
TDK’s latest announcement also flows in this vein of value; specifically, via the creation of more optimal applications in industrial and highly constrained environments, for IoT wearables, automotive, medical, leisure and sports/fitness activity monitoring, select smart home appliances, and for mobile.
TDK has announced the availability of the first automated ML platform integration for the Arm Keil MDK from Qeexo, a TDK company. Qeexo, developers of ML solutions that generate critical and actionable insights from sensor data, helms this brand new Qeexo AutoML platform (which can be tried for free here).
The platform supports a wide range of ML algorithms and is designed for lightweight Cortex-M0 to -M4 class processors with ultra-low latency and power consumption. The Qeexo AutoML platform allows TDK’s customers to smartly leverage sensor data and build and deploy ML solutions rapidly. And, with a very small memory footprint, the platform integration seamlessly supports streamlined, end-to-end embedded ML development workflows.
A few technical details from TDK include (but aren’t limited to):
- The integration encapsulates the ML model into the Arm Keil IDE using the CMSIS-Pack mechanism for running the final custom binary application on an Arm Cortex based MCU. (Arm Keil MDK, by the way, is a comprehensive software development solution for Arm-based microcontrollers. It includes all necessary components for creating, building, and debugging embedded applications.)
- Qeexo AutoML provides a no-code environment, enabling data collection and the training of different ML algorithms (i.e. neural and non-neural networks) to the same dataset.
- Accuracy, memory size, and latency metrics are generated, allowing users to pick the model that best suits their requirements.
“This is all about intuitive innovation,” said Reinhard Keil, Senior Director of Embedded Technology, Arm Keil MDK. “As ML becomes increasingly prevalent, it’s critical that we empower embedded software developers. By abstracting the ML development process with a powerful and easy-to-use graphical user interface, Qeexo AutoML enables IoT and embedded developers to harness the power of ML as they build new solutions.”
“Qeexo AutoML’s integration with Arm Keil MDK closes the gap between embedded and ML development,” said Sang Won Lee, CEO of Qeexo. “This integration enables a full end-to-end, automated, no-code platform for application development. With it, engineers can now collect data, train ML models, validate model, and easily transfer the model to complete the solution development.”
Edited by Greg Tavarez