A normalizing flow learns the post-selection conditional distribution through simulated selection events, then transforms inference back to the pre-selection distribution to correct selection bias.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ME 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Flexible Selective Inference with Flow-based Transport Maps
A normalizing flow learns the post-selection conditional distribution through simulated selection events, then transforms inference back to the pre-selection distribution to correct selection bias.