Comparative neural networks are a new kind of neural networks that can be used to compare two or more items given a set of context features. They compare two items at a time indicating the one that matches the context features better. Consequently, any sorting algorithm, coupled with such a neural comparator, can sort any set of items. Although applications include ink segmentation (for handwriting recognition purposes) and web page ranking, our emphasis on this paper is on classifier aggregation and, in particular, the integration of our standard handwriting recognizers with a user personalization database that consists of user samples.