Research

Publications & Preprints

Note: Some titles were intentionally replaced for double-blind peer reviews.

• Learning to ignore: Single Source Domain Generalization via Oracle Regularization   (NeurIPS 2023, Causal Representation Learning Workshop Poster) [paper][poster]

• ShERPA: Leveraging Neuron Alignment for Knowledge-preserving Fine-tuning   (ICLR 2024, Workshop on Mathematical and Empirical Understanding of Foundation Models (ME-FoMo)) [paper][poster]

• Is Data Augmentation Reliable for Generalization?   *(Working Paper)