Statistical Pattern Recognition Review and Outlook.
Abstract
In spite of the vigorous efforts by researchers, statistical pattern recognition remains to be a field which is rich in challenging theoretical problems. In this paper, emphasis is placed on near future and future outlook of this field while examining the recent development of the field. Ten problem areas where the solutions are most wanted are listed as: feature extraction, nonstationary patterns, adaptive systems, learning complexity, finite sample size effects, computational recognition complexity, contextual analysis, optimum pattern recognizer, statistical and syntactic mixed model, and the automatic generation of recognition rules for complex patterns. These problem areas are closely examined and possible future approaches are suggested. It is pointed out that any 'breakthrough' in solving these problems would probably require fertilization from other fields such as artificial intelligence, digital signal processing, etc.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 25, 1975
- Accession Number
- ADA013615
Entities
People
- Chia‐Hung Chen
Organizations
- University of Massachusetts Dartmouth