Imagine …preventing a crime while it’s still happening, saving a life through remote-control surgery, sensing fires or car accidents in time to avoid them? Thanks largely to Hollywood, we’ve all bought into the value of such time and life saving abilities. But the technology to make them ubiquitous is still largely science fiction because the AI and machine learning (ML) capabilities that make them possible are locked in the cloud.
Extending AI/ML to the network’s edge reduces the latency required for real-time response, but today’s edge lacks the resources to deliver the mandatory ubiquitous availability and 100% reliability. Truly unleashing the power of AI/ML for mission-critical applications requires AI that can learn and predict at the edge.
In this talk, we’ll discuss emerging technologies in learning and prediction at the edge on various form factors such as low power GPUs, FPGAs, mobile CPUs and microcontrollers such as Raspberry Pi/Arduino.----