Previous |  Up |  Next


pre-processing; decomposition; pattern classifiers
In this paper we present a novel approach to decomposing high dimensional spaces using a multiobjective genetic algorithm for identifying (near-)optimal subspaces for hierarchical classification. This strategy of pre-processing the data and explicitly optimising the partitions for subsequent mapping onto a hierarchical classifier is found to both reduce the learning complexity and the classification time with no degradation in overall classification error rate. Results of partitioning pattern spaces are presented and compared with various algorithms.
[1] Friedman J. H.: A recursive partitioning decision role for nonparametric classification. IEEE Trans. Comput. 26 (1997), 4, 404–408
[2] Henrichon E. G., Fu K. S.: A nonparametric partitioning procedure for pattern classification. IEEE Trans. Comput. 18 (1969), 7, 614–624 DOI 10.1109/T-C.1969.222728
[3] Kanal L. N.: Problem–solving models and search strategies for pattern recognition, IEEE Trans. Pattern Analysis Machine Intelligence 1 (1979), 2, 193–201 DOI 10.1109/TPAMI.1979.4766905
[4] Kumar R.: Feature Selection, Representation and Classification in Vision. Ph.D. Thesis, Dept. Electronic and Electrical Engineering, University of Sheffield, 1997
[5] al C. C. Taylor et: Dataset descriptions and results. In: Machine Learning, Neural and Statistical Classification (D. Michie, D. J. Spiegelhalter and C. C. Taylor, eds.), Ellis Horwood, London 1994, pp. 131–174
Partner of
EuDML logo