Interesting direction but the 98.8% FPR in Table 1 seems like a dealbreaker. Anyone understand what's going on with the contradictory results between the text and tables?
dwattttt•1h ago
> Empirically, CTVP attains very good detection rates with reliable false positives
A novel use of the word "reliable"? Jokes aside, either they mean the FPR as the opposite of what you'd expect, the table is not representative of their approach, or they're just... really optimistic?
causalmodels•1h ago
dwattttt•1h ago
A novel use of the word "reliable"? Jokes aside, either they mean the FPR as the opposite of what you'd expect, the table is not representative of their approach, or they're just... really optimistic?