A multiple classifier approach for multisensor data fusion

In many applications of pattern recognition and automated identification, it is not uncommon for data obtained from different sensors monitoring a physical phenomenon to provide complimentary information. In such applications, data fusion-a suitable combination of the complimentary information-can offer more insight into the phenomenon than any of the individual data sources. We have previously introduced Learn⁺⁺, an ensemble based approach, as an effective automated classification algorithm that is capable of learning incrementally. Recognizing the conceptual similarity between data fusion and incremental learning, our approach is then to employ an ensemble of classifiers generated by using all of the data sources available, and strategically combine their outputs. We have observed that the prediction ability of such a system was significantly and consistently better than that of a decision based on a single data source across several benchmark and real world databases.