Table 5.

Performance on the HD benchmark

Method	AUC	CE
sup-CK	0.710 ± 0.19	0.47
sup-CKL	0.752 ± 0.16	0.38
sup-CK-Vol	0.722 ± 0.18	0.46
sup-CKL-Vol	0.766 ± 0.17	0.38
Vol	0.648 ± 0.15	0.89
Princ-Axis	0.650 ± 0.18	0.71
sup-PI	0.702 ± 0.19	0.47
MultiBind	0.69 ± 0.14	0.48
Sequence	0.577 ± 0.09	0.83

Performance for each algorithm is evaluated by its mean AUC score and by its classification error (CE), averaged over all pockets.