A Comparison of the Delta Method and the Bootstrap in Deep Learning Classification
This work provides a faster alternative for uncertainty quantification in deep learning classification, though it is incremental as it compares existing methods on standard datasets.
The paper compared the Delta method and Bootstrap for predictive epistemic uncertainty in deep learning classification, finding a strong linear relationship between their uncertainty estimates and a fivefold computation time reduction with the Delta method.
We validate the recently introduced deep learning classification adapted Delta method by a comparison with the classical Bootstrap. We show that there is a strong linear relationship between the quantified predictive epistemic uncertainty levels obtained from the two methods when applied on two LeNet-based neural network classifiers using the MNIST and CIFAR-10 datasets. Furthermore, we demonstrate that the Delta method offers a five times computation time reduction compared to the Bootstrap.