Table 2.

Maximum relevance-minimum redundancy terms.

Term	Abbreviation	Formula	Interpretation
Dependency	D	D=(1/|S|)∑i∈SI(h, i)	Discriminating features are highly correlated with groups, and the most relevant features of the selection and categorization variables are selected. That is, the feature can reflect information relating to the groups to the greatest extent
Redundancy	R	R=(1/|S|2)∑i,j∈SI(i, j)	Description of the dependency relationship between discriminative features. Minimal relevance between each discriminative feature is required; that is, the principle of minimum redundancy
Mutual information difference	MID	(D − R)	Difference between the maximum relevance and minimum redundancy, represented by the two optimization conditions

I refers to the mutual information value between two features (i and j). D refers to the mutual information value between the discriminative feature and the category (groups). h refers to the groups of the dataset. |S| refers to the number of feature sets. R refers to the redundancy between the features.