A. Tveit. Norwegian University of Science and Technology, IDI/NTNU, N-7491 Trondheim, Norway, MSc Thesis, (1997)
Abstract
The most used method of finding logical rules from
data, inductive logic programming (ILP), has shown
successful, but unfortunately not very scalable with
increasing problem size. In this report a model for
doing induction of logical rules, using the concepts of
the potentially more scalable method of genetic
algorithm, is suggested.
Five strategies of reducing the search space in the
representation are suggested: pruning by logical
entailment, pruning by integrity constraints, pruning
by logic factorisation, pruning by range restriction,
and pruning using a heuristic fitness function on the
cohesion of literals. The genetic operators suggested
are applying these pruning search strategies.
The model has yet to be implemented and tried out in an
experimental setting.
%0 Thesis
%1 GILP1997Tveit
%A Tveit, Amund
%C IDI/NTNU, N-7491 Trondheim, Norway
%D 1997
%K ILP algorithms, genetic inductive logic programming,
%T Genetic Inductive Logic Programming
%U http://amundtveit.info/publications/1997/MScThesisAbstract.php
%X The most used method of finding logical rules from
data, inductive logic programming (ILP), has shown
successful, but unfortunately not very scalable with
increasing problem size. In this report a model for
doing induction of logical rules, using the concepts of
the potentially more scalable method of genetic
algorithm, is suggested.
Five strategies of reducing the search space in the
representation are suggested: pruning by logical
entailment, pruning by integrity constraints, pruning
by logic factorisation, pruning by range restriction,
and pruning using a heuristic fitness function on the
cohesion of literals. The genetic operators suggested
are applying these pruning search strategies.
The model has yet to be implemented and tried out in an
experimental setting.
@mastersthesis{GILP1997Tveit,
abstract = {The most used method of finding logical rules from
data, inductive logic programming (ILP), has shown
successful, but unfortunately not very scalable with
increasing problem size. In this report a model for
doing induction of logical rules, using the concepts of
the potentially more scalable method of genetic
algorithm, is suggested.
Five strategies of reducing the search space in the
representation are suggested: pruning by logical
entailment, pruning by integrity constraints, pruning
by logic factorisation, pruning by range restriction,
and pruning using a heuristic fitness function on the
cohesion of literals. The genetic operators suggested
are applying these pruning search strategies.
The model has yet to be implemented and tried out in an
experimental setting.},
added-at = {2008-06-19T17:35:00.000+0200},
address = {IDI/NTNU, N-7491 Trondheim, Norway},
author = {Tveit, Amund},
biburl = {https://www.bibsonomy.org/bibtex/2fb8fd3b324fd309247a8c2aba9d37d33/brazovayeye},
email = {ae@amundtveit.info},
interhash = {ab09b62bc59198a8f2b8bfc8e9563886},
intrahash = {fb8fd3b324fd309247a8c2aba9d37d33},
keywords = {ILP algorithms, genetic inductive logic programming,},
notes = {Related work:
http://amundtveit.info/publications/1997/MScThesisAbstract.php#CitedBy},
school = {Norwegian University of Science and Technology},
size = {179KB},
timestamp = {2008-06-19T17:53:23.000+0200},
title = {Genetic Inductive Logic Programming},
type = {MSc Thesis},
url = {http://amundtveit.info/publications/1997/MScThesisAbstract.php},
year = 1997
}