One paradigm for nanoelectronics involves building transistor networks and transistor and memory components from thin films containing random or ordered networks of molecules, carbon nanotubes, nanowires, nanoparticles, and organic films. Components and links formed in such networks can show variety of useful non-linear effects. A recent example is the "memristor" (memory transistor) promoted by Hewlett-Packard. In this talk we will describe a number of recent theoretical and experimental approaches to achieve digital and analog computational functionality in such networks, and briefly describe of our recent research toward neuromorphic computing.