Bayesian inference using spintronic technology: A proposal for an MRAM-based stochastic logic gate

摘要: Bayesian inference forms the basis for many models of human cognition as well in development autonomous systems that must rapidly make decisions based upon ambiguous or incomplete sensory data. Recent research has shown a simple Muller C-element can implement Bayes' rule directly using stochastic computation. In this work, design uses spin-based magnetic tunnel junction devices is described. These nanodevices allow compact implementation memory buffers, which are required to reshuffle output bitstream order minimize problem autocorrelation inherent C gate. A rotating buffer be effective mitigating impact when used with multi-input element. Improved area efficiency expected compared previous designs control signals shared across several C-elements. Circuit simulations and numerical analyses presented demonstrate operation