Is combining useful for dissimilarity representations ?

For learning purposes, representations of real world objects can be built by using the concept of dissimilarity. In such a case, an object is characterized in a relative way, i.e. by its dissimilarities to a set of the selected prototypes. Such dissimilarity representations are found to be more practical for some pattern recognition problems. When experts cannot decide for a single dissimilarity measure, a number of them may be studied in parallel. Now the question arises how to make use of all the information given. We investigate two possibilities of combining either dissimilarity representations themselves or classifiers built on each of them separately. Our experiments conducted on a handwritten digit set demonstrate that when the dissimilarity representations are different in nature, a much better performance can be obtained by their combination than on individual representations.