Fast algorithm for feature extraction
Grzegorz Borowik, J. Jankowski, Karol J. Kowalski · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2015
A number of methods for data reduction have already been proposed in the literature. However, these methods, particularly those based on discernibility matrices, are insufficient for large databases and thus Big Data processing. The purpose of this paper is the introduction of a new efficient feature extraction algorithm that is based on statistics of attribute values. Tests have shown that the implementation is much more efficient than currently available solutions.