An analog continuous-time neural network is described. Building blocks
which include the capability for on-chip learning and an example netw
ork are described and test results are presented. We are using analog
nonvolatile CMOS floating-gate memories for storage of the neural weig
hts. The floating-gate memories are programmed by illuminating the ent
ire chip with ultraviolet light. The subthreshold operation of the CMO
S transistor in analog VLSI has a very low power dissipation which can
be utilized to build larger computational systems, e.g., neural netwo
rks. The experimental results show that the floating-gate memories are
promising, and that the building blocks are operating as separate uni
ts; however, especially the time constants involved in the computation
s of the continuous-time analog neural network should be studied furth
er.