Towards Polyhedral Automatic Differentiation
Abstract
Most Automatic Differentiation (AD) tools lack a way to explicitly represent or differentiate performance-critical and hardware-dependent constructs such as parallelism, vectorisation, or tiling. Machine-learning frameworks work around this by hiding implementation details from the AD process, but lack the generality of general-purpose programming languages. Instead, this talk discusses the polyhedral model as a way for general-purpose AD tools to preserve performance tweaks through the differentiation process.
Cite
Text
Hückelheim and Kukreja. "Towards Polyhedral Automatic Differentiation." NeurIPS 2019 Workshops: Program_Transformations, 2019.Markdown
[Hückelheim and Kukreja. "Towards Polyhedral Automatic Differentiation." NeurIPS 2019 Workshops: Program_Transformations, 2019.](https://mlanthology.org/neuripsw/2019/huckelheim2019neuripsw-polyhedral/)BibTeX
@inproceedings{huckelheim2019neuripsw-polyhedral,
title = {{Towards Polyhedral Automatic Differentiation}},
author = {Hückelheim, Jan and Kukreja, Navjot},
booktitle = {NeurIPS 2019 Workshops: Program_Transformations},
year = {2019},
url = {https://mlanthology.org/neuripsw/2019/huckelheim2019neuripsw-polyhedral/}
}