M. T. Pham, K. Tachibana, E.S.M. Hitzer, S. Buchholz,
T. Yoshikawa, T. Furuhashi
Robust geometric feature extraction and classification
Proceedings of ISAMPE 2008, Busan, South Korea, 9-11 Oct. 2008, 8 pages.
Abstract: This research proposes to use geometric algebra to systematically extract geometric features from data given in a vector space. We show the results of classification of hand-written digits, which were classified by feature extraction with the proposed method. Given a set of spatial vectors, we extract k-vectors (scalars, vectors, area elements, ...) of different subspace dimensions k; which encode the variations of the features. Our results confirm that the strategy to mix different geometric algebra feature extractions is superior in both classification precision and robustness (e.g. against random rotations) when compared with pure coordinate value features, which is the most often used conventional method.
[ PDF ] 191K