Random optimized geometric ensembles

We use a single-hidden layer feedforward neural network (SLFN) to interpret the model of optimized geometric ensembles (OGE). Based on the SLFN, we simplify OGE into random optimized geometric ensembles (ROGE), which may contain much less hidden nodes than that of OGE. Furthermore, on 12 UCI data sets we verify that ROGE can achieve the same level of classification performance as OGE in less consumption of space and time.