Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Polynomial approximation of spectral data in LVQ and relevance learning

: Melchert, Friedrich; Seiffert, Udo; Biehl, Michael

Volltext (PDF; )

Machine learning reports 9 (2015), Nr.3, S.25-32
ISSN: 1865-3960
Workshop "New Challenges in Neural Computation" (NC2) <6, 2015, Aachen>
Zeitschriftenaufsatz, Konferenzbeitrag, Elektronische Publikation
Fraunhofer IFF ()

High dimensional data serves as input for a variety of classification tasks. In the case of spectral information, this data can be understood as discrete sampling of an (unknown) underlying function. In this paper we discuss an approach that improves classification performance for spectral data by expanding the data in terms of basis functions. Two real world spectral data classification problems demonstrate the advantages of the method.