The Hopfield neural network has two important aspects. The first is a thresholding amplification scheme. In the simplest situation this can be an electronic amplifier whose output does not go beyond a certain level even if the input to the amplifier is increased. The second aspect characteristic of the network is that of being able to interconnect the outputs of all amplifiers with the inputs of all amplifiers through paths which have specified but selectable resistances to the flow of electrons. The research addresses this latter problem. Photo-sensitive material will be used to program these interconnections to enable the network to be changed at will and show the full flexibility of the Hopfield network.
