A comparative study of some multiple expert recognition strategies

Multiple expert decision combination strategies are being widely used nowadays in various disciplines including pattern recognition and document understanding. These approaches have been proven to be more efficient and robust than their stand-alone counterparts. There is however a popular belief that these multiple expert approaches, although capable of outperforming individual experts, tend to be very costly in terms of computational speed requirements. In this paper, a comparative study of seven established multiple expert decision combination approaches as applied to the recognition of numeric characters is carried out and it is shown that, contrary to popular belief as far as the computational cost is concerned, some of these approaches are extremely competitive with respect to stand-alone experts