Evolving networks to have intelligence realized at nanoscale
Celestine Lawrence · 2018
In this thesis, we have made the following progress. Introducing ‘neuroevolution in nanomaterio’ as a method to reach ultimate physical limits of computing. A mathematically compact treatment of SET networks that enables analytical computation of stability diagrams. A novel mean-field approximation to simulate large scale SET networks. Illustrative simulations that show the rich physics of SET networks that are exploitable for logic, memory and pattern recognition. The first experimental demonstration of Boolean logic functionality in a disordered system at the nanoscale. Conception of the NP internet architecture for neuroevolution, and an illustration of it implementing classification of binary images. A definition of intelligence and an IQ metric, demonstrating its applicability on a nanomaterial cluster of dopant atoms.