Brian Hoskins As Moore’s Law comes to its end and advances in computing shift the market’s demand for computing towards Artificial Intelligence, entirely new architectures, neuromorphic architectures, are coming into vogue. The different requirements of neuromorphic computers, including enormous demand for memory and a high tolerance for defects, is increasingly causing a reassessment of research priorities into nanotechnology and integrated circuit manufacturing. We will review the mathematical foundations of the most important new computing approaches in A.I. and discuss the unique ways these mathematical operations can be accelerated using nanotechnology and novel computing architectures for critical use cases, but especially for online training of neural networks as well as inference in the field where neural networks will increasingly be deployed. Looking forward, we will identify the key challenges that need to be resolved to bring emerging technologies into use as well as future trends in A.I. and computer science that may further drive research into materials science, nanotechnology, and IC design.
Brian Hoskins is a Physicist in the Alternative Computing Group at the National Institute of Standards and Technology. His work primarily focuses on heterogenous integration of emerging technologies with CMOS to improve the performance of novel computing architectures as well as on novel computing approaches to exploit these new technologies. He has a B.S. and M.S. in Materials Science and Engineering from Carnegie Mellon University and a Ph.D. in Materials Science from The University of California, Santa Barbara. He was previously and National Research Council Postdoctoral Fellow at NIST.