Maximum size of the subset to select in each random trial.
Maximum size of the subset to select in each random trial. This inherently defines the maximal size of the cluster that can be detected. Because the algorithm always only evaluates a subset of the features, it will never find correlated clusters of a size bigger than k. Nevertheless, setting k to high leads to a low specificity of the contrast measure.
Defines how many random trials are used to fill the tables.
Defines how many random trials are used to fill the tables. The more tries we allow the more accurate the results will be and the smaller correlations we will find. Can be static or dependent on the number of features.
number of features in the dataset
Defines how many operations are allowed to be performed in parallel
Defines how many operations are allowed to be performed in parallel
Calculate the parameters, esp. number of random tries according to given beta and m.
alpha defines the probability of missing a cluster of size m during the evaluation.