Complex network dynamics in self-assembled atomic-switch networks: prospects for neuromorphic computation

Abstract

The inherent power of the biological brain, with regard to pattern recognition, is unparalleled and cannot even be matched by multi-million dollar supercomputers. Inspired from this, neuromorphic computation, where ideas originating from the complex structure and functionality of the biological brain are utilized for advanced computation has shown great potential. In this regard, we are developing on-chip pattern classification capabilities via inexpensive self-assembly of nanoparticles (NPs). The formation of percolating microstructure of Sn NPs and tunnel junctions leads to a complex atomic-switch network (ASN) poised near criticality. Voltage stimulation is utilized for modulating the synaptic structure of the network, which shows potential for utilization as a ‘reservoir’ in reservoir computing (RC).