Can Networks of Synthetic Neurons Scale to Biological Dimensions?
Recent developments in compact and ultra-low-power digital and analog circuits that emulate key neural building blocks lend credence to the claim that networks of such neurons can reach biological scale in the near future. This talk will highlight some recent advances on this front, including our own efforts to evoke complex biological behaviors with extremely simple digital neurons. However, attempts to integrate these building blocks into networks with biological dimensions must also properly account for the extremely high-radix connectivity of biological neurons. An analysis of recent proposals as well as technology projections for emerging electrical and photonic interconnect technologies reveals a dramatic efficiency gap between biological interconnects and known synthetic approaches. The talk will close with some possible directions to investigate for closing this gap to enable future deployment of large-scale networks of synthetic neurons.
Mikko Lipasti is currently the Philip Dunham Reed Professor of Electrical and Computer Engineering at the University of Wisconsin-Madison, where he leads the PHARM research team. He earned his BS in Computer Engineering from Valparaiso University in 1991, his M.S. in Electrical and Computer Engineering from Carnegie Mellon University in 1992, followed by his Ph.D. in 1997, both under Dr. John Shen. Before and after his Ph.D. work, he learned to ply his craft during his years at IBM, where he helped develop software and hardware for PowerPC servers. He joined Wisconsin in Fall 1999, was granted tenure in 2005, and was promoted to Full Professor in 2009. He has consulted for Intel Corporation and (briefly) for Sun Microsystems. His primary research interests include high-performance, low-power, and reliable processor cores; networks-on-chip for many-core processors; and fundamentally new, biologically-inspired models of computation.