search

luferrer / DCA-PLDA

Discriminative Condition-Aware PLDA
Other
42 stars 7 forks source link

Question when using compute_performance_with_confidence_intervals #9

Closed JaesungHuh closed 1 month ago

JaesungHuh commented 2 months ago

Hi, i think there is a bug in the code below. https://github.com/luferrer/DCA-PLDA/blob/e47aff3c336128942e866ece81838d4ec99d7237/dca_plda/scores.py#L484

The original code is :

sel_key = Key(np.array(key.test_ids)[sel_test_idxs], np.array(key.enroll_ids)[sel_enroll_idxs], key.mask[np.ix_(sel_enroll_idxs,sel_test_idxs)])

but according to the init function of Key, I think it should be :

sel_key = Key(np.array(key.enroll_ids)[sel_enroll_idxs], np.array(key.test_ids)[sel_test_idxs], key.mask[np.ix_(sel_enroll_idxs,sel_test_idxs)])

Please let me know if I'm wrong.

Thank you.

Best wishes, Jaesung

luferrer commented 1 month ago

You are completely right. I have pushed the fix. Thanks for letting me know.