From data centers, through edge accelerators to endpoint devices: Artificial intelligence (AI) Applications range from large scale analysis of medical data and online retail recommendation engines, to robotics and computer vision, to sensor fusion in the tiniest sensor nodes. The infusion of AI techniques into so many areas of computing is changing compute paradigms across the board.
Our Virtual Event will provide answers to questions like: How to keep up with these changes, especially given AI’s propensity to evolve at a staggering rate? How does one design chips or systems for a constantly shifting workload like this? How does one make the call between maximising performance today and keeping some flexibility for the sake of future-proofing?
AI in the Data Center
AI in the data center is revolutionising online retail in the cloud and applications like medical imaging and the financial sector at the enterprise level. What are the challenges of AI compute at this scale? And are the big CPU and GPU players keeping up with challenges from a new breed of high-profile AI chip startups?
AI at the Edge
Advances in neural networks and specialised accelerator hardware are making more and more AI possible outside the data center. Edge compute “boxes” in the field as well as device-level applications like robotics are becoming more and more powerful, processing more data more quickly. But what are the limits of what can be done at the edge, and where are the opportunities to save power and cost?
AI in the Device
While AI techniques traditionally required huge compute power, the technology has come far enough that even battery-powered devices can take advantage of it. Under what circumstances does it make sense to analyse sensor data where it is collected, at the periphery of the IoT? The growing trend for voice activation and voice control of consumer electronics and smart home appliances is a big driver for AI in endpoint devices, and can even be done with microcontrollers today. This event will explore what’s possible with the bare minimum of power.