Pith. sign in

Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

citation-role summary

method 1

citation-polarity summary

fields

cs.LG 1

years

2025 1

verdicts

CONDITIONAL 1

roles

method 1

polarities

use method 1

representative citing papers

Thumb on the Scale: Optimal Loss Weighting in Last Layer Retraining

cs.LG · 2025-06-24 · conditional · novelty 6.0

For square-loss weighted ERM in the proportional asymptotic regime, the class weight that equalizes per-class errors is ρ̃ = π−/π+ + (π−/π+ − 1) δ/(2π+ − δ), exceeding the ratio of priors and growing with δ.

citing papers explorer

Showing 1 of 1 citing paper.

  • Thumb on the Scale: Optimal Loss Weighting in Last Layer Retraining cs.LG · 2025-06-24 · conditional · none · ref 1

    For square-loss weighted ERM in the proportional asymptotic regime, the class weight that equalizes per-class errors is ρ̃ = π−/π+ + (π−/π+ − 1) δ/(2π+ − δ), exceeding the ratio of priors and growing with δ.