Analog convolution neural networks for energy-efficient real-time signal processing.
-
Value
$1,798,755
-
Prompt Contribution
$792,802
This project aims to develop low-power AI hardware for battery- or ambient-powered smart sensors. Using transformative analog electronics techniques, we aim to offer AI solutions within strict power limits, suitable for sectors such as smart cities, industrial Internet of Things, technologies for agriculture, and surveillance/security.
In close collaboration with Irréversible, our team will design and manufacture innovative analog modules to demonstrate the potential of our analog memory technology for next-generation AI. This collaboration will enable us to validate the operation of the analog resistive memories developed at 3iT in an audio detection application.
This technology is also central to Irréversible’s products and underpins its R&D program, which is part of Sherbrooke’s Quantum Innovation Zone ecosystem, DistriQ, with a particular focus on AI-assisted quantum computing applications. This project therefore fosters the development of disruptive technological solutions and will reinforce Quebec’s role in hardware innovation for low-energy AI.
Organisation (s)
