“…Table 2 compares the performance of different classifiers with the low labeling ratio of 10%. We use V for vehicle, S for speed, SMRC for supervised MRC, LR for logistic regression, NB for naive Bayes, KSVM for kernel SVM, LP for label propagation, SSMRC for semi-supervised MRC and LP-W for label propagation with a semi-supervised weighting algorithm in (2). We see that when the labeling ratio is low, supervised MRC performs poorly, label propagation works well, and semi-supervised MRC works the best.…”