The learning sequence with the Neural ID Cure algorithm begins with the digitization of “good” examples. Features from the resulting scenes get extracted via ordinary feature extraction routines. The resulting information feeds to the Cure engine which builds influence areas for each of the “good” feature examples. To classify unknown parts, the Cure algorithm compares features from the unknown to the corresponding influence region of “good” features to look for a match. It then uses the resulting information to make a decision about the part. For illustration purposes, shown here is a decision made by simply summing the outputs of the feature partitions. If the sum exceeds a certain number, Cure decides the unknown part is “good.” Farmore- complicated decision-making schemes are possible if some of the unknown features lie outside the corresponding influence region. Influence regions can also be deployed in layers to form a hierarchy if necessary to handle complex decisions.