A Flexible Artificial Synapse for Neuromorphic System

Neuromorphic computing, as a new paradigm, highlighted for its highly parallel, energy efficient features, has attracted a lot of attention. The hardware implementation for a neuromorphic system proposes the strong desire for suitable building blocks. The synaptic device is a very promising solution because of its stimulation-history-related response, which fits the nature of a neural network. In this work, an artificial synapse based on a memristive transistor fabricated by a simple process is realized. The device not only shows multi-level states which is the main feature of a memristor and is essential to hardware implementation neuromorphic system, but also exhibits physical flexibility, a feature that supports wearable and portable electronics. On this basis, a proof-of-feasibility simulation using the experimental data is performed to realize the pattern classification.