Neuromorphic behavior of memristor models.

Ponente(s): Eduardo Elpidio Rodríguez Martínez, Juan Gonzalo Barajas Ramírez

Memristors are two terminal theoretical devices that obey an internal-variable dependent Ohm's law, that results in a resistive memory effect. Using these devices electronic circuits can be designed to mimic the physiological behaviours of biological neural systems. Two important features of biological neurons are the learning mechanisms called Long-Term Potentiation (LTP) and habituation (HAB), which are important for the learning process as they result in strengthening or weakening of synaptic connections between neurons. In this contribution we use memristors to mimic synapses that perform the LTP and HAB learning phenomena by applying a train of repeated impulses. These results provide a large range of new possibilities to implement artificial neural networks and learning rules.