LED-7 is the task of identifying the digit on a standard 7-LED digital display, with
approximately 10% of the features
ipped (to simulate random noise). Due to the
noise, the optimal probability of a correct classication is about 74%. LED-24 is the
same task with an additional 17 irrelevant features that serve as distractors. The
data sets used for these tasks each have 250 instances. Monks-2 is an articial data
set with 6 features, where the class description to be learned is exactly two features
have the value 1." It has 432 test instances, of which 169 are designated as training
cases. Lymph is a set of 159 medical cases provided by Zwitter and Soklic. The task is
to predict a medical diagnosis given 18 descriptive features. Promoters is a set of 106
E. coli DNA sequences, where the task is to predict whether the sequence will act as
a promoter. Soybean is a collection of crop disease records. The task is to predict the
disease given 35 features providing information about the growing conditions. This
task has 307 designated training instances, and 376 designated test instances