Analytical and Interactive Techniques for Multivariate Data Compression and Classification
Abstract
The report demonstrates the applicability of classical statistical techniques to problems involving compression and classification of multivariate data. The theoretical foundations of two such techniques, intrinsic analysis and discriminant analysis, are treated in detail. Efficient digital computer implementation is discussed, including the combined application of intrinsic and discriminant analysis and a new algorithm for computing approximate intrinsic bases for very large problems. Experimental results are presented on the application of these techniques as feature extractors in a signal classification problem. Also included is a description of the interactive graphics-oriented system software which has been developed to facilitate the application of these techniques.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 25, 1970
- Accession Number
- AD0719753
Entities
People
- Robert B. Roper