Hardware implementation of classical conditioning with iron-oxide-based memristors

摘要

Memristors have attracted considerable interest for the hardware implementation of brain-inspired neuromorphic devices and systems. In this work, we achieved an emulation of classical conditioning, which is a general ability of the brain to link information together, using iron oxide memristors. Utilizing the tunable conductance and set/reset threshold characteristics, we developed a simple neuromorphic circuit with one single memristor, which can provide history-dependent synaptic weight. After training with a specific temporal pairing of conditioned and unconditioned stimuli, the memristor could be tuned from the high-resistance state to the low-resistance state, and consequently, the conditioned response was successfully realized. (C) 2018 The Japan Society of Applied Physics