Hayashi et al. describe an innovative use of singular value decomposition (SVD) techniques to explore feature extraction and gesture recognition in this proceedings paper. It provides a nice presentation with respect to SVD decomposition and extracting features in time series, as described in its abstract, by decomposing left singular vectors, which provides representations of patterns of motion. This leads to left singular vectors that affect the results of the matrix representation.
The paper consists of an abstract, categories and subject descriptors, general terms, and keywords. The paper then consists of the following sections: “Introduction,” “Feature Extraction Using SVD,” and “Gesture Recognition.” It concludes with an acknowledgement and three references.
This is a well-done, excellent, and creative paper and is quite interesting as it combines aspects of SVD methods and technology with gesture recognition, including a description of experimental measurements and a brief table of experimental results.