LSI ranks training samples by informativeness using the KL divergence between Laplace-approximated posteriors with and without each sample, and the ordering transfers from a small probe to larger models.
Algorithmic stability for adaptive data analysis
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Laplace Sample Information: Data Informativeness Through a Bayesian Lens
LSI ranks training samples by informativeness using the KL divergence between Laplace-approximated posteriors with and without each sample, and the ordering transfers from a small probe to larger models.