DocumentCode :
3218127
Title :
Kolmogorov´s structure functions with an application to the foundations of model selection
Author :
Vereshchagin, Nikolai ; Vitányi, Paul
Author_Institution :
Dept. of Math. Logic & Theor. of Algorithms, Moscow State Univ., Russia
fYear :
2002
fDate :
2002
Firstpage :
751
Lastpage :
760
Abstract :
Kolmogorov (1974) proposed a non-probabilistic approach to statistics, an individual combinatorial relation between the data and its model. We vindicate, for the first time, the rightness of the original "structure function", proposed by Kolmogorov: minimizing the data-to-model code length (finding the ML estimator or MDL estimator), in a class of contemplated models of prescribed maximal (Kolmogorov) complexity, always results in a model of best fit (expressed as minimal randomness deficiency). We show that both the structure function and the minimum randomness deficiency function can assume all shapes over their full domain (improving an old result of L.A. Levin and both an old and a recent one of VV Vyugin). We determine the (un)computability properties of the various functions and "algorithmic sufficient statistic.".
Keywords :
computational complexity; probability; Kolmogorov structure functions; MDL estimator; ML estimator; combinatorial relation; data-to-model code length; maximal complexity; minimum randomness; model selection; structure function; Engineering profession; Logic; Mathematical model; Mathematics; Maximum likelihood estimation; Proposals; Region 8; Shape; Statistics; Technological innovation;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Foundations of Computer Science, 2002. Proceedings. The 43rd Annual IEEE Symposium on
ISSN :
0272-5428
Print_ISBN :
0-7695-1822-2
Type :
conf
DOI :
10.1109/SFCS.2002.1182000
Filename :
1182000
Link To Document :
بازگشت