Classification Improvement by Optimal Dimensionality Reduction when Training Sets are of Small Size.
Abstract
When the sizes of the training sets are small, classification in a subspace of the original data space may give rise to a smaller probability of error than the classification in the data space itself. This is because the gain in the accuracy of estimation of the likelihood functions used in classification in the lower dimensional space (subspace) offsets the loss of information associated with dimensionality reduction (feature extraction). To test this conjecture, a computer simulation was performed. A number of pseudo-random training and data vectors were generated from two four-dimensional Gaussian classes.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1976
- Accession Number
- ADA032672
Entities
People
- D. L. Van Rooy
- R. J. P. Defigueiredo
- S. A. Starks
Organizations
- Rice University