Abstract

We propose a novel hypercube view that per-ceives the label space of multi-label classifi-cation problems geometrically. The view al-lows us to not only unify many existing multi-label classification approaches, but also de-sign a novel algorithm, Principle Label Space Transformation (PLST), which seeks impor-tant correlations between labels before learn-ing. The simple and efficient PLST relies on only singular value decomposition as the key step. Experimental results demonstrate that PLST is faster than the traditional Bi-nary Relevance approach and is superior to the modern Compressive Sensing approach in terms of both performance and efficiency. 1.