Knowledge representations evolvable using GALE.
This package contain some of the knowledge representations than can be evolved
using the GALE model. In this distribution
three different representations are provided:
- Rules (using sorted decision lists)
- Instances (nearest neighbor classification)
- Decision trees
The rules implement rules using the n-ari alphabet defined over {0,1,...,n#}. No
real-value representation is provided in this distribution (see decision trees).
Instances are prototypes build on the data using nearest neighbor rule. This
knowledge representation can be applied indistinctly to data with nominal and
numeric attributes. Finally, the last knowledge representation distributed is
based on orthogonal decision trees. These trees deal with nominal and numeric
data. Hence, rule for numeric attributes can be extracted for the evolved
decision trees.