Varberg, Sweden, December 20, 2022; Ekkono Solutions AB, a global pioneer in edge machine learning software, has been granted €2.5M in funding from the EIC (European Innovation Council). The investment will be used to accelerate the product and market development of federated learning to supplement Ekkono’s software suite.
“Federated learning is a superhot segment in machine learning today”, says Rikard König, CTO at Ekkono. “We bring federated learning to a whole new level. Our edge machine learning solution continuously trains an individual model per device, which enables personalized condition monitoring, predictive maintenance, and performance optimization. With federated learning you can learn from what many products have learnt individually. Our solution is designed for IoT and provides new insights that solve additional use cases for smart products.”
Federated Learning is a machine learning technique that enables decentralized devices to learn from each other without sharing data. Each device trains its own model on local data, related to its own configuration and environment. By sending the learnings of each device to the cloud, they can be combined to distribute the learning among the devices. To date, most federated learning research has been focused on enabling the learning from millions of smartphones. Ekkono works with industrial equipment, enabling machine learning on constrained microcontrollers, to learn where, how, and for what the individual device is used. With Ekkono’s federated learning solution, models from a fleet of devices can be compared to identify units that act irrational compared to its peers, and be combined to generate even better models.
“Synthesis, as our federated learning product is called, was already on our roadmap”, says Jon Lindén, CEO of Ekkono. “The EIC funding accelerates the development, putting it sooner in the hands of our customers, and further expanding our technology advantage. Our main challenge will be to find and hire the people we need for this.”
“We’re thankful for the support we’ve received from our partners and customers on this, including the six strategic customers who signed letters of intent expressing their interest in Synthesis”, continues Jon Lindén. “And Zaz Ventures who supported us through the application process.”
For more information:
Jon Linden, CEO, Ekkono Solutions AB, ph. +46-709-576 006
Rikard König, CTO, ph. +46-73-922 96 56
Ekkono in short
Ekkono Solutions AB is a Swedish software company that does edge machine learning, i.e. enables machine learning to run onboard physical devices. Ekkono’s unique capabilities, which is the result of over seven years of research at the University of Borås, enable product OEMs to rapidly develop customized smart features that are self-learning, predictive and sustainable. Ekkono addresses multiple verticals, including industrial equipment, automotive, and energy, and has a strong portfolio of tier-1 customers. www.ekkono.ai