A Parzen-window classifier architecture for massively-integrated nanoscale resonant devices

As a proposal for the concept of computing using quantum phenomena, we are presenting a system architecture best suited to massively-integrated nano-scale resonant devices. It is a Parzen window classifier which determines the class label of a given input using a number of learned sample data stored in nano devices. The main idea is collecting small bumps made from nano-scale devices having Gaussian-like current-voltage characteristics. A multi-dimensional bump is realized by summation and thresholding of output currents. An analog Parzen window classifier was designed and the concept has been demonstrated by SPICE simulation. In the design the resonance characteristics of nano devices were emulated by simple NMOS bump circuits.