Instar and outstar learning with memristive nanodevices

Abstract

The instar and outstar synaptic models are among the oldest and most useful in the field ofneural networks. In this paper we show how to approximate the behavior of instar andoutstar synapses in neuromorphic electronic systems using memristive nanodevicesand spiking neurons. Memristive nanodevices are especially attractive for thisapplication since such devices are tiny, can be densely packed in crossbar-likestructures and possess the long time constants, or memory, needed by the synapticmodels.