Title :
On two nonequivalent measures of complexity
Author :
Ramirez, Donald E.
Author_Institution :
Dept. of Math., Virginia Univ., Charlottesville, VA, USA
Abstract :
A careful distinction is made between the two definitions of complexity introduced by M.H van Emden (1971) for data compression. It is shown that the two concepts are nonequivalent, and a compromise criterion is introduced. Inherently different properties of the two concepts of complexity are proved. It is noted that these differences are crucial to the choice of a performance criterion in model building. The decision rule of M.S. Maklad and S.T. Nichols (1980) is explicitly derived in the case of a multivariate linear regression model
Keywords :
data compression; modelling; complexity; data compression; decision rule; model building; multivariate linear regression model; performance criterion; Control theory; Covariance matrix; Data compression; Econometrics; Eigenvalues and eigenfunctions; Entropy; Linear regression; Maximum likelihood estimation; Parameter estimation; Statistical analysis;
Journal_Title :
Systems, Man and Cybernetics, IEEE Transactions on
