Pattern Recognition: Human and MechanicalThe first major work in the nascent discipline of ``cognitive science.'' It provides a unified presentation of pattern recognition that introduces new mechanical methods as well as a wider humanistic perspective on the science. Showing that practically all the known pattern recognition algorithms can be derived from the principle of minimum entropy, it provides the first complete theory of pattern recognition. |
Contents
PATTERN RECOGNITION | 1 |
PATTERN RECOGNITION AS PERCEPTION | 21 |
PATTERN RECOGNITION AS CATEGORIZATION | 45 |
Copyright | |
13 other sections not shown
Common terms and phrases
algorithm assume atoms Bayesian becomes belonging called centers of mass classification clustering coefficients components concept conditional probability considered convergence corresponding covariance matrix curve deductive defined density function determined disjunction distribution distributive law domain eigenvalue eigenvectors entropy equation estimate evaluation example extra-evidential fact factor factor analysis Figure finite number formula Fourier transform G-matrix given grammatical Hence heuristic human hyperplane hypothesis i-th idea inductive ambiguity infinite introduce lattice LMSE method logical machine mathematical matrix maximum likelihood means memory minimization minimum entropy n-dimensional neighbor method neurons normal normal distribution object obtain orthogonal paradigmatic symbol paradigms pattern recognition perceptron picture plane position possible predicates principle prior probability problem relation represents result rules sample satisfy sentence similarity structure subspace Suppose symmetry theorem theory usually variables vector zero