In many practical situations it is not feasible to collect labeled samples for all available classes in a domain. Especially in supervised classification of remotely sensed images it is impossible to collect ground truth information over large geographic regions for all thematic classes. As a result often analysts collect labels for aggregate classes (e.g., Forest, Agriculture, Urban). In this paper we present a novel learning scheme that automatically learns sub-classes (e.g., Hardwood, Conifer) from the user given aggregate classes. We model each aggregate class as finite Gaussian mixture instead of classical assumption of unimodal Gaussian per class. The number of components in each finite Gaussian mixture are automatically estimated. A semi-supervised learning is then used to recognize subclasses by utilizing very few labeled samples per each sub-class and a large number of unlabeled samples. Experimental results on real remotely sensed image classification showed not only improved accuracy in aggregate class classification but the proposed method also recognized sub-classes accurately.