We present one application of the measure of maximum entropy for credal sets: as a total uncertainty measure to branch classification trees based on imprecise probabilities. In this paper we justify the use of maximum entropy as a global uncertainty measure for credal sets. A deduction of this measure, based on the best lower expectation of the logarithmic score is presented. We have also carried out several experiments in which credal classification trees are built taking a global uncertainty measure as basis. The results show that the error is lower when the maximum entropy is used as global uncertainty measure.
Keywords. Imprecise probabilities, uncertainty, maximum entropy, imprecision, non-specificity, classification,classification trees, credal sets
Paper Download
The paper is availabe in the following formats:
Authors addresses:
Joaquín Abellán
Dpto. Ciencias de la Computación
ETSI Informática
18071 Granada
SPAIN
Serafín Moral
Dpto. Ciencias de la Computación e IA
ETSI Informática
Universidad de Granada
18071 Granada - Spain
E-mail addresses:
| Joaquín Abellán | jabemu@teleline.es |
| Serafín Moral | smc@decsai.ugr.es |
Related Web Sites