Thumb on the Scale: Optimal Loss Weighting in Last Layer Retraining
Published in NeurIPS 2025, 2025
Optimal class loss weighting in last layer retraining
Publications
Selected Work
Label Noise Robustness for Domain-Agnostic Fair Corrections via Nearest Neighbors Label Spreading
Published in NeurIPS 2024, 2024
Cleaning labels to improve robustness of two-stage corrections
Theoretical Guarantees of Data Augmented Last Layer Retraining Methods
Published in ISIT 2024, 2024
Theoretical comparison of several popular WGA correction methods for LLR
Robustness to Subpopulation Shift with Domain Label Noise via Regularized Annotation of Domains
Published in TMLR, 2024
Robust worst-group accuracy via biased pseudo-annotation
Smoothly Giving Up: Robustness for Simple Models
Published in AISTATS 2023, 2023
Boosting with $\alpha$-loss and its application to simple models