Biased accumulation based on multiplexer using stochastic correlated logic

Abstract

AbstractStochastic computing is a non‐traditional computing approach that utilizes stochastic bitstreams to represent numerical values. This representation enables the implementation of circuits using basic logic gates, named stochastic circuits. This work proposes a versatile architecture with correlated input bitstreams for biased accumulation. The design offers a notable reduction in the reliance on random number generators, resulting in lower hardware costs and higher computing accuracy when compared with the existing designs. Experimental results show that the application of Gaussian smoothing has a reduction in area by at least 45% and an improvement in peak signal‐to‐noise ratio of at least 30%.