Forest-RC

A similar method as Forest-RI to construct Random Forest or $k$ decision trees but here new attributes are generated by linear combinations of existing attributes to reduce Correlation between the individual classifiers. At each node attributes are randomy selected.