Data Augmentation

Is augmentation effective in improving prediction in imbalanced datasets?