Relevant Component Analysis (RCA) tries to find a linear transformation of the feature space such that the effect of irrelevant variability is reduced in the transformed space.