OPERATIONS RESEARCH ON RECOGNITION. THE GENERALIZED SEARCH PROCESS
Abstract
Techniques for detection, analysis, and description of unknown recurrent signals are examined. Distributional search techniques (based on analysis of the empirical probability distribution of the input) are shown to have efficiences comparable to an unrealizable maximum-likelihood estimator, but become fairly complicated whenA LARGE AMOUNT OF INFORMATION ABOUT THE UNKNOWN SIGNAL IS REQUIRED. Linear search techniques, exemplified by various types of adaptive filters, are examined analytically; results of a diagnostic adaptive filter simulation program are discussed. Linear search techniques yield finegrain information about unknown signals more readily than the distributional techniques, but are subject to certain intrinsic limitations. Several alternative means for implementation of adaptive filters are presented, and the operational characteristics of a configuration employing cascaded adaptive filters are discussed. Hybrid techniques and a novel search technique based on certain transformation properties of probability distributions are discussed. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1961
- Accession Number
- AD0266874
Entities
People
- Robert D. Turner
Organizations
- General Electric