Beyond Von Neumann Systems

Computer systems have made rapid progress over the past sixty years. However, there has been very little change in the Von Neumann style principle for designing system architecture. While this approach has worked extremely well for number crunching and data manipulation applications, a drastically different architectural model and computational approach may be required for allowing machines to reach the cognitive abilities of the human brain for perceptual tasks. This talk will introduce brain-inspired system architectures that employ brain-inspired algorithms on traditional computational fabrics, machine learning accelerators and neuromorphic architectures that mimic functional fabrics in the brain. Next, the talk will introduce systems that exploit the intrinsic physics of a dynamically coupled oscillator system to solve computationally complex tasks efficiently. The seminar will provide new insights on the interplay between new device technologies and computational models.