New Interfaces in Neural Computing
Friday, September 22, 2017 — 8:30AM - 9:15AM
“Technology is the active human interface with the material world.” -UK LeGuin
Computer science is the art of reducing complex, messy real-world problems to tractable abstractions. The layers of abstraction developed by computer scientists over the last fifty years have formed a diverse ecosystem of high level paradigms to create, understand, and distribute information with speed, reliability, and efficiency.
Neuroscience today is done at what we could consider the “transistor level,’’ but the questions neuroscientists face are often identical to those asked by current frontiers of computing. These questions include ``How should data be represented, and how do we decide what to keep?’’ or “How much power does it cost to build our network, and can we do better?’’ In this talk, we will explore the parallels between problems involving power efficiency, redundancy, and structure between distributed computing and biologically plausible neural-networks, and in this translation demonstrate the power of an interdisciplinary approach to neural computing.
Avani Wildani, Emory University