When running a typical logistic regression model using klqp I get good intercept and some coefficients. The smaller coefficients are always inflated. This does not improve with more input data. Is there another inference method in the lib that would do this better (Other than hmc)? Like maybe the exact classes?
Related Topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
ed.KLqp VI seems to have trouble with small parameters with sparse data
|
0 | 607 | April 1, 2018 | |
Why my KLqp cannot infer good parameters on simple iris regression task
|
0 | 827 | January 24, 2018 | |
Variational inference with KLqp for LSTM example
|
0 | 841 | February 25, 2018 | |
Implicit model inferences example
|
0 | 677 | September 6, 2018 | |
Biased variance estimates when using KLqp for Bayesian Linear Regression
|
7 | 1290 | February 9, 2018 |