Embedded AI: why bother and what is the problem?
Posted on 11/08/2017
AI, Neural Networks or Machine Learning technologies are being adopted in many circumstances, but usually all the data has to be sent to a server for processing before being acted upon.
There would be many advantages of being able to do this analysis on the embedded system itself – faster decision making and the ability to implement AI in situations where the data link is not possible, but the problem is that traditional embedded compute platforms cannot handle the processing power required. The solution is not just faster silicon – but new architectures and ways to implement systems.
This free event on the 7th September in London, will bring together technology suppliers and technology users to discuss the opportunities and barriers to the deployment of next generation systems and there will be presentations from experts (industry and academic) as well as current and potential users who will present their knowledge, promote discussion and share learning.
So that the focus is on the key technology issues you can help by completing the following brief survey: https://www.surveymonkey.co.uk/r/SmartSilicon
In the survey you can register your interest in the event, giving a pitch session or propose speakers – so please complete the survey.
Registration details for the 7th September event will be released next week on: https://ktn-uk.co.uk/events
The event will be called “Embedded AI Part 1- Smart Silicon
(Note: – Part 2 of this series will focus on applications of Embedded AI).
Contact Nigel Rix for more details.