- This event has passed.
Challenges and Opportunities for Efficient & Scalable Neuromorphic Systems
March 9 @ 10:30 am - 11:59 am
Co-sponsored by: Sudip Shekhar
The emerging Internet of Everything (IoE) era will demand cyberphysical systems that continuously interact with the physical sensory world and are capable of information processing, learning, cognition, inference and automated action, much like the living “brain-body-environment” interactive systems in nature. These cyberphysical systems need to be extremely energy efficient, scalable in size from mm-scale at the edge of IoE to km-scale in the backend cloud datacenters, and universal in their capabilities to support a variety of existing and emerging functions seamlessly. The fundamental neurosynaptic architecture of the brain demonstrates such phenomenal efficiency and scalability in nature. We will discuss the challenges and opportunities for building neuromorphic systems that emulate some of the useful functional mechanisms of the brain by utilizing Moore’s Law scaling, and CMOS as well as beyond-CMOS technologies. Going well beyond evolutionary Von Neumann computing and machine learning, the neuromorphic computing systems of the future will feature efficient & sparse spatio-temporal information encoding, fully distributed memory & compute, scalable “small world” networks, structural & synaptic plasticity, asynchronous & event-driven operation, continuous & distributed learning, autonomous power management including homeostasis, stochasticity, adaptivity, variation tolerance & resilience.
Speaker(s): Vivek De,
Electrical & Computer Engineering
2332 Main Mall
Vancouver, British Columbia