Predicting Resource Requirement in Intermediate Palomar Transient Factory Workflow
CCGrid, 2020
This paper proposes a Bayesian-network model that exploits the spatiotemporal correlation of astronomical images to predict per-stage resource requirements in transient-detection pipelines like iPTF, and shows it identifies the most influential spatial and temporal features and achieves prediction errors close to the intrinsic random-variance limit.
BibTeX
@inproceedings{kang2020predicting,
title={Predicting Resource Requirement in Intermediate Palomar Transient Factory Workflow},
author={Kang, Qiao and Sim, Alex and Nugent, Peter and Lee, Sunwoo and Liao, Wei-keng and Agrawal, Ankit and Choudhary, Alok and Wu, Kesheng},
booktitle={2020 20th IEEE/ACM International Symposium on Cluster, Cloud and Internet Computing (CCGRID)},
pages={619--628},
year={2020},
organization={IEEE}
}