Unbiased Gradient Estimation in Unrolled Computation Graphs with Persistent Evolution Strategies (Extended Abstract)
Abstract
Current approaches for optimizing parameters in unrolled computation graphs suffer from high variance gradients, bias, slow updates, or large memory usage. We introduce a method called Persistent Evolution Strategies (PES), which divides the computation graph into a series of truncated unrolls, and performs an evolution strategies-based update step after each unroll. PES eliminates bias from these truncations by accumulating correction terms over the entire sequence of unrolls. PES allows for rapid parameter updates, has low memory usage, is unbiased, and has reasonable variance.
Cite
Text
Vicol et al. "Unbiased Gradient Estimation in Unrolled Computation Graphs with Persistent Evolution Strategies (Extended Abstract)." International Joint Conference on Artificial Intelligence, 2022. doi:10.24963/IJCAI.2022/750Markdown
[Vicol et al. "Unbiased Gradient Estimation in Unrolled Computation Graphs with Persistent Evolution Strategies (Extended Abstract)." International Joint Conference on Artificial Intelligence, 2022.](https://mlanthology.org/ijcai/2022/vicol2022ijcai-unbiased/) doi:10.24963/IJCAI.2022/750BibTeX
@inproceedings{vicol2022ijcai-unbiased,
title = {{Unbiased Gradient Estimation in Unrolled Computation Graphs with Persistent Evolution Strategies (Extended Abstract)}},
author = {Vicol, Paul and Metz, Luke and Sohl-Dickstein, Jascha},
booktitle = {International Joint Conference on Artificial Intelligence},
year = {2022},
pages = {5354-5358},
doi = {10.24963/IJCAI.2022/750},
url = {https://mlanthology.org/ijcai/2022/vicol2022ijcai-unbiased/}
}