Methods for Automatic Classification in Body Surface Mapping
Our problem may be stated as follows: a digital computer receives map data for several classes of heart diseases and has to decide for a new case into which class this case should be filed. This is a well-known problem and solved in general for pattern classification: A map is described by a finite number n of variables, called features. A particular map is considered as a point in an n-dimensional pattern space encompassing the region in which patterns can occur. Sets of patterns are classified to pattern classes. Then the problem is to classify a set of unknown maps into one of the known classes.
KeywordsDecision Boundary Pattern Classification Automatic Classification Pattern Class Intrinsic Dimensionality
Unable to display preview. Download preview PDF.
- Kanal, L. N., and Chandrasekaran, B., 1968, On Dimensionality and Sample Size in Statistical Pattern Classification, Proc. NEC Conf., pp. 2–7.Google Scholar
- Meisel, W. S., 1972, Computer-oriented Approaches to Pattern Recognition, Acad. Press. N.Y., pp. 12–15.Google Scholar
- Postweiler, W., 1979, Diplomarbeit, Universitat Karlsruhe.Google Scholar
- Sammon, J. D., Foley, D., and Proctor, A., 1970, Considerations of Dimensionality versus Sample Size, IEEE Symp. Adaptive Processes, Austin, Texas.Google Scholar