Neuromorphic computing systems based on flexible organic electronics

Abstract Today software systems known as neural networks are at the basis of numerous artificial intelligence applications and are successfully implemented to translate languages, classify images, recognize diseases, and form the basis of the spur in autonomous driving. However, these algorithms require a substantial amount of computer resources and energy. The brain on the other hand, operates in a highly parallel fashion, connecting neurons via synapses, rendering it compact and highly efficient in recognizing patterns, speech, and images. Neuromorphic engineering takes advantage of the efficiency of the brain by mimicking and implementing essential concepts such as neurons and synapses in hardware. In this chapter we review the development of organic neuromorphic devices. We highlight efforts to mimic essential brain functions, such as spiking phenomena, spatiotemporal processing, homeostasis, and functional connectivity and demonstrate related applications. Next, we review important metrics for implementing low-power and reliable neuromorphic computing, such as state retention and conductance tuning. Finally, we give an outlook on future directions and potential applications, with a particular focus on interfacing with biological environments.