Optimal Learning Recognizer for Unknown Signal Sets in a Channel with Feedback Link
A general Kalman filer type supervised optimum quadric dichotomizer for an on-off sequence of unknown signal is presented. Then, the optimum analogue feedback is obtained which makes the average of forward signal transmission energy minimum. The extension of the proposed machine to nonsupervised problems is also discussed. The effectiveness of a feedback method over a non-feedback method is clarified. It is also tested by computer simulation.
KeywordsTransmission Energy Unknown Signal Feedback Method Feedback Link Message Probability
Unable to display preview. Download preview PDF.
- (2).C.V. Jakowatz, R.L. Shuey, and G.M. White, “Adaptive waveform recognition,” in Proc. 4th Symp. on Information Theory, C. Cherry, Ed., Washington, D.C.: Butterworths, 1961.Google Scholar
- (4).K. Tanaka, S. Higuchi, and S. Maekawa, “An identification method of system characteristics using a new type of adaptive correlating filter,” in Proc. Int. Fed. Automatic Control Tokyo Symp., Aug. 1965, pp. 245–254.Google Scholar
- (17).J.S. Meditch, Stochastic Optimal Linear Estimation and Control. New York: McGraw-Hill, 1969.Google Scholar
- (18).K. Tanaka and S. Tamura,”Some considerations on a type of pattern recognition using nonsupervised learning procedure,” in Int Fed, Automatic Control int, Sympt Technical and Biological Problems of Control, Yerevan, Armenia, Sept. 1968.Google Scholar
- (19).K. Tanaka, “Some studies on pattern recognition with non- supervised learning procedures,” in Pattern Recognition and Machine Learning, K.S. Fu Ed., New York: Plenum, 1971, pp. 1–17.Google Scholar
- (20).J.C. Hancock and P.A. Wintz, Signal Detection Theory, New York: McGraw-Hill, 1966.Google Scholar