Multiple classifier systems for multisensor data fusion

We have previously introduced Learn⁺⁺, an ensemble of classifiers based algorithm capable of incremental learning from additional data, and pointed to its feasibility in data fusion applications. In this contribution, we provide additional details, updated results and insight on how such a system can be used in integrating complementary knowledge provided by different data sources obtained from different sensors. Essentially, the algorithm generates an ensemble of classifiers using data from each source, and combines these classifiers using a weighted voting procedure. The weights are determined based on the individual classifier's training performance as well as the observed or predicted reliability of each data source.