Uncertainty Quantification with the Empirical Neural Tangent Kernel

Joseph Wilson, Chris van der Heide, Liam Hodgkinson, Fred Roosta

Keywords: Neural Tangent Kernel, Bayesian Deep Learning, Uncertainty Quantification, Neural Network

TL;DR: Scalable, post-hoc, Bayesian uncertainty quantification method using an ensemble of linearised networks.

Abstract: While neural networks have demonstrated impressive performance across various tasks, accurately quantifying uncertainty in their predictions is essential to ensure their trustworthiness and enable widespread adoption in critical systems. Several Bayesian uncertainty quantification (UQ) methods exist that are either cheap or reliable, but not both. We propose a post-hoc, sampling-based UQ method for overparameterized networks at the end of training. Our approach constructs efficient and meaningful deep ensembles by employing a (stochastic) gradient-descent sampling process on appropriately linearized networks. We demonstrate that our method effectively approximates the posterior of a Gaussian Process using the empirical Neural Tangent Kernel. Through a series of numerical experiments, we show that our method not only outperforms competing approaches in computational efficiency--often reducing costs by multiple factors--but also maintains state-of-the-art performance across a variety of UQ metrics for both regression and classification tasks.

Primary Area: Probabilistic methods (e.g., variational inference, causal inference, Gaussian processes)

Submission Number: 11152