Table 11.

Averaged and truncated confusion matrix for multi classification using the Valencian-Cohen dataset for multi-class classification with 300 test images (extended-sized test dataset), for 10 runs, using 40/40/20 percent of imbalance setting (for SSDL). From left to right, using $n_l=70$, and $n_l=90$ labels respectively. From top to bottom, the supervised model (with completely balanced labels), the SSDL model with no PBC, and the SSDL model with the PBC.