Figure 1.

Performance of our classifiers, measured on the GXD dataset according to the different performance metrics, calculated over the various cross-validation settings. NB denotes Naïve Bayes; RF denotes Random Forest classifier. The suffix 5 denotes average over 5 complete runs of 5-fold cross validation (25 runs in total); the suffix 10 denotes average over10 complete runs of 10-fold cross validation (100 runs in total). half-training-halftesting represents runs in which half of the GXD dataset was used for training and the other half for testing.