Tokyo University researchers create energy-efficient AI edge IoT devices using MRAM architecture and TGBNN algorithm.

Researchers from the Tokyo University of Science have developed a Magnetic RAM (MRAM)-based architecture that enhances AI capabilities for edge IoT devices. Utilizing a new training algorithm called Ternary Gradient BNN (TGBNN), this design reduces circuit size and power consumption while maintaining performance. The innovation promises efficient AI in applications like wearable health monitors and smart homes, contributing to sustainability by lowering energy usage.

5 months ago
5 Articles