Stochastic computing with spiking neural P systems

This paper presents a new computational framework to address the challenges in deeply scaled technologies by implementing stochastic computing (sc) using the spiking neural p (sn p) systems. sc is well known for its high fault tolerance and its ability to compute complex mathematical operations using minimal amount of resources. however,one of the key issues for sc is data correlation. this computation can be abstracted and elegantly modeled by using sn p systems where the stochastic bit-stream can be generated through the neurons spiking. furthermore,since sn p systems are not affected by data correlations,this effectively mitigate the accuracy issue in the ordinary sc circuitry. a new stochastic scaled addition realized using sn p systems is reported at the end of this paper. though the work is still at the early stage of investigation,we believe this study will provide insights to future ic design development. © j.ucs.