Convergence of the Edited Nearest Neighbor,
Abstract
The edited k-nearest neighbor rule (k-NNR) consists of eliminating those samples from the data which are not classified correctly by the k-NNR and the remainder of the data and using the NNR with the samples which remain from to classify new observations. Wilson has shown that this rule has an asymptotic probability of error which is better than the k-NNR. A key step in his development is showing the convergence of the edited nearest neighbor. His lengthy argument is replaced here by a somewhat simpler one which uses an intuitive fact about the editing procedure. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1973
- Accession Number
- AD0763839
Entities
People
- T. J. Wagner
Organizations
- University of Texas at Austin