Scalable inference for crossed random effects models
Abstract/Contents
- Abstract
- Crossed random effects models are widely used in statistics for modeling data with crossed structure with applications ranging from political science to recommendation systems. However, the computational costs of estimation and inference of the crossed random effects models limit its application on large-scale data sets. With N observations, the cost of computing maximum likelihood estimates scales as O(N^{3/2}) at best. A similar situation arises for Bayesian inference, where the Gibbs sampler takes O(N^{1/2}) iterations to converge, leading once again to a total cost of O(N^{3/2}). In this talk, we propose a variant of backfitting algorithm in the context of likelihood based inference to compute a generalized least squares estimate each iteration of which costs O(N) and provably converges in O(1) iterations. We illustrate our method on a data set from Stitch Fix. We shall also briefly discuss provably-scalable variants of standard approaches besides likelihood based inference.
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource. |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2022; ©2022 |
| Publication date | 2022; 2022 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Ghosh, Swarnadip |
|---|---|
| Degree supervisor | Hastie, Trevor |
| Degree supervisor | Owen, Art B |
| Thesis advisor | Hastie, Trevor |
| Thesis advisor | Owen, Art B |
| Thesis advisor | Athey, Susan |
| Thesis advisor | Linderman, Scott |
| Degree committee member | Athey, Susan |
| Degree committee member | Linderman, Scott |
| Associated with | Stanford University, Department of Statistics |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Swarnadip Ghosh. |
|---|---|
| Note | Submitted to the Department of Statistics. |
| Thesis | Thesis Ph.D. Stanford University 2022. |
| Location | https://purl.stanford.edu/vb765vv9433 |
Access conditions
- Copyright
- © 2022 by Swarnadip Ghosh
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...