Reservoir computing is a new computational framework based on recurrent neural networks. Researchers have developed a new artificial synapse of alpha-indium selenide (α-In2Se3) that may aid in replicating biological neural processes in neuromorphic devices.
Reservoir computing uses artificial synapses to run deep learning algorithms directly without needing data transfer between a memory and a processing unit. α-In2Se3 has a variety of beneficial optoelectronic, ferroelectric, and semiconducting properties.
α-In2Se3 is a very interesting material and a good reservoir computing platform. Its rich physical properties enable the development of multimode and multiscale reservoir computing systems, which will broaden the application scenarios of physical reservoir computing.
α-In2Se3 has two intriguing inherent physical properties: ferroelectric switching and optoelectronic response. Researchers created a planar device that uses in-plane ferroelectric polarizations for electrical synapse while incorporating light as a third terminal to enable optoelectronic response. This one-of-a-kind structure combines the two physical properties and can take advantage of ferroelectric and optoelectronic coupling for heterosynaptic plasticity and higher-level computing functionalities.
Electrical and optical stimuli can be used to control the optoelectronic synapse. It means it will eventually be able to mimic the brain’s innate plasticity while also directly processing information. The researchers discovered that it performed well in handwritten digit recognition and QR code recognition tasks.
Related Content: 2D Perovskites Improve Optoelectronic Devices