Detecting Underspecification with Local Ensembles
This work addresses the issue of model reliability for practitioners by detecting when models are underspecified, though it appears incremental as it builds on existing ensemble and Hessian-based techniques.
The authors tackled the problem of detecting underspecification in pre-trained models by introducing local ensembles, a method that uses local second-order information to approximate prediction variance across model ensembles. They demonstrated its capability in detecting underspecified models on test data, with applications to out-of-distribution detection, spurious correlates, and active learning.
We present local ensembles, a method for detecting underspecification -- when many possible predictors are consistent with the training data and model class -- at test time in a pre-trained model. Our method uses local second-order information to approximate the variance of predictions across an ensemble of models from the same class. We compute this approximation by estimating the norm of the component of a test point's gradient that aligns with the low-curvature directions of the Hessian, and provide a tractable method for estimating this quantity. Experimentally, we show that our method is capable of detecting when a pre-trained model is underspecified on test data, with applications to out-of-distribution detection, detecting spurious correlates, and active learning.