Fractionally-Supervised Classification
Irene Vrbik, Paul D. McNicholas · 2016
Traditionally, there are three species of classification: unsupervised, supervised, and semi-supervised. Supervised and semi-supervised classification differ by whether or not weight is given to unlabelled observations in the classification procedure. In un-supervised classification, or clustering, either there are no labelled observations or the labels are ignored altogether. A priori it can very difficult to choose the optimal level of supervision, and the consequences of a sub-optimal choice can be rather severe. A flexible fractionally-supervised approach to classification is introduced, where any level of supervision — ranging from unsupervised to supervised — can be attained. Our approach uses a weighted likelihood, wherein weights control the level of supervision. Gaussian mixture models are used as a vehicle to illustrate our fractionally-supervised classification approach; however, it is broadly applicable and variations on the pos-tulated model can easily be made by adjusting the weights. A comparison between our approach and the traditional species is presented using benchmark model-based clustering data.