Organic electronics Axon-Hillock neuromorphic circuit: towards biologically compatible, and physically flexible, integrate-and-fire spiking neural networks
Mohammad Javad Hosseini, Elisa Donati, Tomoyuki Yokota, Sunghoon Lee, Giacomo Indiveri, Takao Someya, Robert A. Nawrocki
Abstract
Abstract Spiking neural networks (SNNs) have emerged as a promising computational paradigm to emulate the features of natural neural tissue physiology. While hardware implementations of SNNs are being conceived to emulate biological systems, they typically rely on hard and rigid silicon electronics that are not bio-compatible. In the physical, or materials realm, organic electronics offer mechanical flexibility and bio-compatibility, allowing for the construction of neural processing systems that can be directly interfaced to biological networks. This study introduces an organic electronics implementation of an Integrate-and-Fire spiking neuron based on the Axon-Hillock CMOS circuit. The circuit employs organic p-type and n-type field effective transistors and reproduces the behavior of the CMOS neuromorphic counterpart. We demonstrate its operating characteristics measuring its spike rate output as a function of its input current. We show how it properly integrates input currents and demonstrate its computing abilities in a basic current summing experiment. The static and dynamic power dissipation is calculated to be less than 0.4 and 40 µ W, respectively. This is the first demonstration of the spiking Axon-Hillock neuromorphic circuit using organic materials.