Invited Keynote at ISCA'19 Unary Computing Workshop
Title: “Unary Computing: The Stochastic Circuit Approach”
Speaker: John P. Hayes
University of Michigan
Ann Arbor, USA
Abstract: The unary computation paradigm is found in a variety of forms involving both deterministic and random data representations. Randomness has advantages in computing, as suggested by its widespread occurrence in nature, ranging from quantum mechanics to the human brain. Stochastic computing (SC) is a (re)emerging computational technique that processes data defined by pseudo-random unary bit-streams. It mimics aspects of the nervous system, and enables complex arithmetic operations to be performed using extremely small, low-power, and error-tolerant circuitry. SC has applications in several important areas such as image processing, complex coding techniques, and the design of artificial neural networks. However, achieving acceptably accurate results is difficult and can require excessively long run times. This talk examines the underlying concepts of SC and its applications, and discusses recent research results that focus on the accuracy issue. Among the major sources of inaccuracy are random fluctuations in individual bit-streams, correlations between bit-streams, and inadequate randomness sources. For example, input bit-streams denoting constant stochastic numbers play an important role in SC but are a significant source of random fluctuation errors. However, it is possible to remove error-inducing constant inputs from stochastic circuits by employing suitable sequential designs. Finally, we will discuss the benefits and limitations of SC in the design of neural networks.
John P. Hayes is Professor of EECS at the University of Michigan, Ann Arbor where he holds the Claude E. Shannon Chair of Engineering Science. He has a B.E. degree in electrical engineering from the National University of Ireland, and M.S. and Ph.D. degrees from the University of Illinois. Prior to joining Michigan, he was a faculty member at the University of Southern California. His teaching and research interests include computer-aided design and testing, computer architecture, and unconventional computing techniques. He is a Fellow of both IEEE and ACM. Professor Hayes has authored over 325 technical papers, several patents, and seven books, including Computer Architecture and Organization (3rd ed., 1998), Quantum Circuit Simulation (2009), and Design, Analysis and Test of Logic Circuits under Uncertainty (2012). He has served as editor of various technical journals, including the Communications of the ACM and the IEEE Transactions on Parallel and Distributed Systems. In 2004, he received the Alexander von Humboldt Foundation’s Research Award. He received the IEEE Lifetime Contribution Medal for outstanding contributions to test technology in 2013, and the ACM Pioneering Achievement Award for contributions to logic design, fault tolerant computing, and testing in 2014.