Modifying iterated Laplace approximations
Tiep Mai, Simon Wilson
TL;DR
This paper introduces modifications to the iterLap method, improving its accuracy and efficiency by adjusting stopping rules, residual functions, starting points, and Hessian scaling, demonstrated through illustrative examples.
Contribution
The paper presents novel modifications to the iterLap approximation method, enhancing its accuracy and computational performance.
Findings
Modified iterLap reduces approximation error.
Trade-offs between accuracy and running time are demonstrated.
New residual function and scaling improve convergence.
Abstract
In this paper, several modifications are introduced to the functional approximation method iterLap to reduce the approximation error, including stopping rule adjustment, proposal of new residual function, starting point selection for numerical optimisation, scaling of Hessian matrix. Illustrative examples are also provided to show the trade-off between running time and accuracy of the original and modified methods.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsGaussian Processes and Bayesian Inference · Neural Networks and Applications · Control Systems and Identification