Engineering Commons^™

Investigating R(T) Functions For Spike-Timing-Dependent Plasticity In Memristive Neural Networks, Farhana Afrin, Kurtis D. Cantley

Electrical and Computer Engineering Faculty Publications and Presentations

Brain-inspired neuromorphic computation can be extremely efficient at very large scales due to inherent parallelism, scalability, and fault and failure tolerance. One widely used, biologically plausible synaptic learning mechanism is spike-timing-dependent plasticity (STDP). The proposed generic model of time-varying resistance, or R(t) elements in this work, can produce classical and beyond classical STDP in electronic spiking neural networks with memristive synapses. Hebbian and Anti-Hebbian STDP is verified with the proposed generic R(t) model by tuning the R(t) function. By appropriately placing R(t) functions with selective resistance values, symmetric or non-classical STDP learning behavior is achieved.