Fig. 1: Binary computing vs. stochastic computing.

a Binary computing. Two example 3-bit binary fraction inputs \(A\) and \(B\), representing 4/8 and 6/8, respectively, are computed to yield a binary multiplication output \(C\) of 3/8. When input \(B\) undergoes a bit-flip and the value is changed from 6/8 to 2/8, the output is altered from 3/8 to 1/8. \({FA}\) is short for Full Adder. b Stochastic computing. Two example 8-bit stochastic number inputs \(a\) and \(b\), with probabilities \(P(a)\) and \(P(b)\) of 4/8 and 6/8, respectively, are computed to yield a stochastic multiplication output \(P(c)\) of 3/8. When input \(b\) undergoes a bit-flip and the value is changed from 6/8 to 5/8, the output remains at 3/8.