Variance reduction properties of the reparameterization trick
- Publisher:
- MLR Research Press
- Publication Type:
- Conference Proceeding
- Citation:
- AISTATS 2019 - 22nd International Conference on Artificial Intelligence and Statistics, 2019, 89, pp. 1-10
- Issue Date:
- 2019
Closed Access
| Filename | Description | Size | |||
|---|---|---|---|---|---|
| 1809.10330v3.pdf | Submitted version | 1.16 MB |
Copyright Clearance Process
- Recently Added
- In Progress
- Closed Access
This item is closed access and not available.
The reparameterization trick is widely used in variational inference as it yields more accurate estimates of the gradient of the variational objective than alternative approaches such as the score function method. Although there is overwhelming empirical evidence in the literature showing its success, there is relatively little research exploring why the reparameterization trick is so effective. We explore this under the idealized assumptions that the variational approximation is a mean-field Gaussian density and that the log of the joint density of the model parameters and the data is a quadratic function that depends on the variational mean. From this, we show that the marginal variances of the reparameterization gradient estimator are smaller than those of the score function gradient estimator. We apply the result of our idealized analysis to real-world examples.
Please use this identifier to cite or link to this item: