A method of multiple classifier fusion for SoftMan's perception system

Artificial Life is the simulation, extension and expansion of natural life. As a kind of software Artificial Life, SoftMan is the virtual robot living and acting in computer network world. It has a humanized structure and simulating human, and its perception system should be able to recognize perceptive objects. A cooperative classification model for SoftMan's perception system has been proposed, in which, different humanized senses are simulated by multiple classifier systems (MCS). An essential issue in the research of MCS is the method of multiple classifier fusion. Accordingly, this paper proposes a method of multiple classifier fusion based on decision profile. Simulated experiments were done to compare the method with the Decision Templates (DT) method. The experimental results show the presented method is effective.