The essential difference between the "machine learning" rough set attribute reduction algorithm and the MRMR algorithm

1. Rough set attribute reduction algorithm only selects the condition that the attribute importance is large, and does not consider the redundancy between the conditional attributes in the reduction, and the resulting reduction is not always necessary, that is, it contains the redundancy attribute.

2. The MRMR algorithm, in addition to considering the correlation between the characteristics and the category, also considers the redundancy between features and features, the maximum correlation between the constraint features and the category, and the minimum redundancy of features and features.

3. Based on the MRMR algorithm, the algorithm is improved to reduce the minimum dependent maximum dependency attribute reduction algorithm as follows

The essential difference between the "machine learning" rough set attribute reduction algorithm and the MRMR algorithm