Fusion of two visual perception systems utilizing cognitive diversity

摘要

Decision-making tasks that are based on perception have been performed routinely by human beings in daily life and decision-makers in daily work. It has been observed that the combination of two perception-based decisions could be better than one, provided that the decision-makers communicate with each other and jointly adopt the most confident judgment; simply select the most confident judgment without any dyadic interaction, or communicate with each other on shareable arguments based on reasoning. In this paper, we report a recently conducted visual perception experiment with decisions from nineteen pairs of subjects and their reported confidence factors when making such a decision. Utilizing the concept of a “cognitive diversity” between two systems and the combinatorial fusion algorithm, our results demonstrated that the fusion of two visual perception decision systems can be better than each of the individual systems only if the two systems perform relatively good and they are cognitively diverse. Our study not only provides a robust algorithm to fuse two visual perception decision-making systems, but also suggests a resilient approach to use the cognitive diversity in fusion when the performance of each individual system is not known or cannot be obtained, which is often the case for complex problems.