Artificial neural networks are intended to be used in future

nanoelectronics since their biological examples seem to be robust to

noise. In this paper, we analyze the robustness of Local Cluster Neural

Networks and determine upper bounds on the mean square error for noise

contaminated weights and inputs.