Application of Automatic Clustering to Emitter Identification
Abstract
The goal of clustering is the partitioning of a given set of objects into subsets called clusters in such a way that the objects in a cluster are similar to one another and that objects in different clusters are dissimilar. Clustering may help in getting a more or less direct understanding of the relationships among the objects, and it may be useful as a first step in pattern recognition. Some possible applications are automatic phoneme recognition, data base management systems, personnel classification, detection of errors in files and computer security. Several clustering methods were applied to data sets of practical importance. Automatic pattern recognition using the k nearest neighbors was applied. An efficient method for selecting a good subset from the full of 44 features was tried. In all cases, the results were good. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1976
- Accession Number
- ADA033916
Entities
People
- James Slagle
- Richard C. Lee
Organizations
- United States Naval Research Laboratory