Skip to main content
. Author manuscript; available in PMC: 2025 Oct 15.
Published in final edited form as: IEEE Access. 2025 Jul 28;13:133351–133369. doi: 10.1109/access.2025.3593420

TABLE 4.

T-test results comparing the ood auc in the (panda,artif) experiment using uni features.

(a) Statistical comparison for vaeabmil
Model OoD/auroc t_stat p_value Significant
vaeabmil 0.9993 ± 0.0004 - - -
daeabmil 0.9988 ± 0.0001 2.2618 0.1521 False
dsmil 0.5408 ± 0.0386 20.3737 0.0024 True
clam 0.6794 ± 0.0276 20.3732 0.0024 True
dftdmil 0.6883 ± 0.0300 18.1359 0.0030 True
transmil 0.6982 ± 0.0410 12.8230 0.0060 True
abmil 0.7546 ± 0.0255 16.7655 0.0035 True
(b) Statistical comparison for daeabmil
Model OoD/auroc t_stat p_value Significant

Model OoD/auroc t_stat p_value Significant
daeabmil 0.9988 ± 0.0001 - - -
vaeabmil 0.9993 ± 0.0004 -2.2618 0.1521 False
dsmil 0.5408 ± 0.0386 20.5614 0.0024 True
clam 0.6794 ± 0.0276 20.0429 0.0025 True
dftdmil 0.6883 ± 0.0300 17.9140 0.0031 True
transmil 0.6982 ± 0.0410 12.6833 0.0062 True
abmil 0.7546 ± 0.0255 16.5333 0.0036 True