Long-term Potentiation in Memristive Neuromorphic Systems

The memristor, a recently developed electronic component, behaves analogously to synapses in biological neural networks. Neuromorphic systems, which model biological neurons as electronic circuits, can implement memristors as synapses. Memristive devices were fabricated at the University of Michigan using tungsten oxide. These devices were to be used in neuromorphic systems, but they did not survive to circuit implementation. Instead, memristive synapses were modeled using a field-programmable gate array (FPGA). These modeled synapses exhibited LTP, and could be replaced by a working memristive device in the circuit.