Implementation of biologically plausible spiking neural network models on the memristor crossbar-based CMOS/nano circuits

Memristor nanodevices have good properties for use as synapses to add dynamic learning to neuromorphic networks implemented in crossbar-based CMOS/Nano hybrids. In this paper, we propose and analyze spike-timing-dependent-plasticity (STDP) rule for memristor crossbar based spiking neuromorphic networks. The learning method is implemented by using CMOS based neurons which generate two-part spikes similar to biological action potentials (APs) and send them to both forward and backward directions along their axon and dendrites, simultaneously. The local learning method can modify the state of nanodevices with regards to pre- and postsynaptic spike timings.