Title :
Adaptive Representations for Pattern Recognition
Author :
Sears, Raymond W., Jr.
Author_Institution :
The Johns Hopkins University, Baltimore, Md.
Abstract :
An adaptive method of generating a set of correlation patterns for measuring the characteristics of unknown patterns is discussed. The correlation patterns change with time to follow slow changes in the configuration of the inputs so that they remain optimum for the most recent inputs. The relationship between the changing set of measurement patterns and the actual identification process is discussed. Results of experiments with a simulated system are presented.
Keywords :
Character generation; Computational modeling; Computer errors; Cost function; Distortion measurement; Helium; Measurement standards; Nonlinear distortion; Pattern recognition; Signal processing;
Journal_Title :
Systems Science and Cybernetics, IEEE Transactions on
DOI :
10.1109/TSSC.1965.300062
