Ekkono's edge machine learning software is the result of 7 yrs of university research, and enables individual learning at the edge, even when scaling down to minimal MCU platforms
Everything gets connected. The key to IoT success is automation. Automation in IoT spells Machine Learning (ML). The traditional approach to ML is that you collect data from thousands of device to find common denominators. There are not that many common denominators in IoT. With Ekkono's edge machine learning we turn things around; By actually training the ML model at the edge, we learn the normal state for the individual machine/vehicle/thing. This way we can predict what will happen, and detect when something changes or deviates from that normal. Deviations from what's expected can be used for condition-based/predictive maintenance. The extension of this is simulation of how to run the device in an optimal and/or sustainable way. This is enabled by Ekkono's Edge Intelligence software!
Besides the fact that we actually make it possible to run machine learning – learning, not just inference – even on constrained devices at the edge of the network, we expedite the development of these self-learning, predictive and smart features, and we enable our customers to harmonize on one solution for edge machine learning.
Edge machine learning enables a plethora of crucial IoT features: Individual health indicators (e.g. ball bearing temperature), virtual sensors, condition-based/predictive maintenance, auto-optimization, auto-installation, smart battery management, and more...