STUDY OF SOLUTION REPRESENTATION LANGUAGE INFLUENCE ON EFFICIENCY OF INTEGER SEQUENCES PREDICTION

Naučno-tehničeskij Vestnik Informacionnyh Tehnologij, Mehaniki i Optiki

关键词: genetic programming;    universal prediction;    integer sequences;    algorithmic complexity;   

DOI  :  10.17586/2226-1494-2015-15-1-107-114

来源: DOAJ

【 摘 要 】

Methods based on genetic programming for the problem solution of integer sequences extrapolation are the subjects for study in the paper. In order to check the hypothesis about the influence of language expression of program representation on the prediction effectiveness, the genetic programming method based on several limited languages for recurrent sequences has been developed. On the single sequence sample the implemented method with the use of more complete language has shown results, significantly better than the results of one of the current methods represented in literature based on artificial neural networks. Analysis of experimental comparison results for the realized method with the usage of different languages has shown that language extension increases the difficulty of consistent patterns search in languages, available for prediction in a simpler language though it makes new sequence classes accessible for prediction. This effect can be reduced but not eliminated completely at language extension by the constructions, which make solutions more compact. Carried out researches have drawn to the conclusion that alone the choice of an adequate language for solution representation is not enough for the full problem solution of integer sequences prediction (and, all the more, universal prediction problem). However, practically applied methods can be received by the usage of genetic programming.

【 授权许可】

Unknown