V2O5 based artificial optical synaptic devices for neuromorphic computing

Abstract

This paper proposed novel V2O5 based optical synaptic devices using wafer scalable microfabrication techniques. These V2O5 optical devices mimic the human brain like functions such as paired pulse facilitation (PPF), short term memory and long term memory. PPF index of 209 % which indicates promising short term memory behavior. The maximum retention time of 3.6 × 103 seconds was achieved. These V2O5 optical devices were used as synapses in an artificial neural network (ANN) to classify digits and alphabets with accuracy of 98.49 % and 97.58 % respectively. Thus, the proposed V2O5 based optical synaptic device integrated for real-world applications can be a breakthrough innovation in recognition that is computationally efficient.