In this day and age of digital living, data is considered to be the new oil. Most of the standard netizens generate a large amount of data everyday. Some of it is sensitive too, and needs to be safeguarded. One of the ways to keep data safe is by processing it on the end machines, and this is exactly what the London-based AI startup Edgify does.
The startup’s offering enables its clients to process data on edge (end) devices. The data can be used to train AI models, without needing to upload any data to the cloud. Edgify has now raised €5.5M in its seed funding round to further develop its services and here are the details.
Funding details
Edgify raised a notable €5.5M funding in its latest seed round funding. With this round, the total amount of funding raised by the company, according to Dealroom, stands at €16.8M. This seed funding round is backed by Octopus Ventures, Mangrove Capital Partners, and an unnamed semiconductor ‘giant’. The startup plans to utilise the funds for further developing its offering, which it refers to as the framework.
Ofri Ben-Porat, CEO and co-founder of Edgify, comments, “Edgify allows companies, from any industry, to train complete deep learning and machine learning models, directly on their own edge devices. This mitigates the need for any data transfer to the cloud and also grants them close to perfect accuracy every time, and without the need to retrain centrally.”
Edgify’s Federated Learning framework
As mentioned before, Edgify offers a framework to help companies process data on edge devices. These edge devices can be any connected devices, and with the help of the startup’s framework, they can process generated data. This processed data can then be used to train a complete AI model locally, without any need to upload the data.
Edgify calls its offering the Federated Learning framework. After training an AI model locally, the learning can also be shared across a network of similar devices. This approach can be used for helping AI subsets such as computer vision, NLP, voice recognition and other forms of AI to better understand their objective.
As for the machines that can be trained using the company’s federated learning network, the list includes virtually any device with a CPU, GPU or NPU (Neural Processing Unit), including MRI machines, connected cars, checkout lanes, and mobile devices, among others. As per the company, the accuracy of an AI model trained on its Framework is averaged at 99.98% and it is claimed to never decrease. Such an approach could help companies mitigate service and hardware costs, along with time required for training AI in a centralised or cloud based facility.
Image credits: Edgify