Sonia Buckley, NIST, USA
Martin Schell, Fraunhofer Institut, Germany
Volker Sorger, George Washington Univ., USA
S.J. Ben Yoo, Univ. of California Davis, USA
What will the future ML and AI Systems look like? Will photonic quantum and neuromorphic computing play a role?
This workshop addresses the future of the machine learning and artificial intelligence systems facing significant challenges from ML related workloads doubling every 3-4 months while Dennard’s law scaling has ceased to keep pace with Moore’s law nearly 15 years ago. Modern data centers and computing systems are now showing intelligence comparable or surpassing the capabilities of human intelligence, but at the cost of extremely high power consumption. On the other hand, neuromorphic accelerators promise 3-6 orders of magnitude improvements in energy-efficiency and throughput compared to the traditional von Neumann computing, and quantum computing claims quantum supremacy to solve nearly unsolvable problems in a record time. What will the future ML and AI systems look like? Will quantum and neuromorphic computing play an active role? What role will photonics play in this context?
The workshop will be organized in two parts.
Part I will address the challenges faced by today’s data centers and computing systems in coping with explosively growing ML and AI workloads. In particular, Part I will discuss future trends and ML and AI computing systems, and covers opportunities for fundamental changes in computing system architectures enabled by intelligent programmable photonics.
Part II will discuss emerging neuromorphic and quantum computing technologies to efficiently and effectively accelerate ML training and inference workloads. In particular, we will address opportunities and challenges of photonics in computing for future hyperscale data centers and computing systems handling AI and machine learning workloads
To be determined.