Computation using mismatch: Neuromorphic extreme learning machines

Abstract

In this paper, we describe a low power neuromorphic machine learner that utilizes device mismatch prevalent in today's VLSI processes to perform a significant part of the computation while a digital back end enables precision in the final output. The particular machine learning algorithm we use is extreme learning machine (ELM). Mismatch in silicon spiking neurons and synapses are used to perform the vector-matrix multiplication (VMM) that forms the first stage of this classifier and is the most computationally intensive. System simulations are presented to evaluate the dependence of performance (in a classification and a regression task) on analog and digital parameters like weight resolution, maximum spike frequency etc. SPICE simulations show that the proposed implementation is ≈ 92X more energy efficient as opposed to custom digital implementations for a classification task with 100 dimensional inputs. Measurement results for a regression task from a field programmable analog array (FPAA) fabricated in 0.35μm CMOS are presented as a proof of concept.