Learning Incoherent Subspaces: Classification via Incoherent Dictionary Learning

In this article we present the supervised iterative projections and rotations (s-ipr) algorithm, a method for learning discriminative incoherent subspaces from data. We derive s-ipr as a supervised extension of our previously proposed iterative projections and rotations (ipr) algorithm for incoherent dictionary learning, and we employ it to learn incoherent sub-spaces that model signals belonging to different classes. We test our method as a feature transform for supervised classification, first by visualising transformed features from a synthetic dataset and from the ‘iris’ dataset, then by using the resulting features in a classification experiment. ∗This work has been supported by the Platform Grant EP/K009559/1 and the Leadership Fellowship EP/G007144/1, both from the UK Engineering and Physical Sciences Research Council (EPSRC).