A Brief Review of Automatic Rule Generation

Abstract

Automatic rule generation can be regarded as rule
generation techniques, and it can be decomposed into
rule induction and rule-base optimization. As the
aspect of the size of the system, it just need to work
rule induction in a simple or a small system with a few
variables in the relative database without processing
the step of rule-base optimization, because the
performance of rule-base optimization is almost
achieved by the step of rule induction in a simple or a
small system. So far as the size of work space, the
family of shared partitions is suitable to work in a
small size work space with a good convergence, and
the family of clustering is appropriate to work in a
large size work space with a good convergence.
However, the family of shared partitions is easy to
make the problem of curse of dimensionality when
dealing with large systems.