|
Activity Number:
|
91
- High Dimensional Data, Causal Inference, Biostats Education, and More
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 9, 2021 : 10:00 AM to 11:50 AM
|
|
Sponsor:
|
SSC (Statistical Society of Canada)
|
|
Abstract #318971
|
|
|
Title:
|
Laplace and Saddlepoint Approximations in High Dimensions
|
|
Author(s):
|
Yanbo Tang* and Nancy Reid
|
|
Companies:
|
University of Toronto and University of Toronto
|
|
Keywords:
|
Higher-Order Asymptotics;
Laplace Approximation ;
Saddlepoint Approxiamtion ;
Approximate Bayesian Computing
|
|
Abstract:
|
The Laplace and saddlepoint approximations are commonly used density approximations in frequentist and Bayesian inference, respectively. Their asymptotic properties are well understood when the number of parameters in the model is held fixed. However, their asymptotic behavior when the number of parameters is allowed to grow with the number of observed data points is not as well studied. In this work we show that for general models, under certain high-dimensional asymptotic regimes, the approximation error is asymptotically negligible for both approximation methods. Some specialized results for the linear exponential family are also considered.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2021 program
|
