Deploying machine learning (ML) models in the wild is challenging as it suffers from
distribution shifts, where the model trained on an original domain cannot generalize well to …

Estimating the out-of-distribution performance in regimes where labels are scarce is critical
to safely deploy foundation models. Recently, it was shown that ensembles of neural …

Model calibration aims to align confidence with prediction correctness. The Cross-Entropy
CE) loss is widely used for calibrator training, which enforces the model to increase …

Unsupervised domain adaptation (UDA) has seen substantial efforts to improve model
accuracy for an unlabeled target domain with the help of a labeled source domain. However …