Scalable computation of prediction intervals for neural networks via matrix sketching. (arXiv:2205.03194v1 [stat.ML])

Web: http://arxiv.org/abs/2205.03194

Accounting for the uncertainty in the predictions of modern neural networks
is a challenging and important task in many domains. Existing algorithms for
uncertainty estimation require modifying the model architecture and training
procedure (e.g., Bayesian neural networks) or dramatically increase the
computational cost of predictions such as approaches based on ensembling. This
work proposes a new algorithm that can be applied to a given trained neural
network and produces approximate prediction intervals. The method is based on
the classical delta …

arxiv computation ml networks neural neural networks prediction scalable