This repository contains codes for the SDP-based frame work to provide robust bound for comparing, quantising and pruning Neural Networks developed by J. Li, R. Drummond, S.R. Duncan in 'Robust error bounds for quantised and pruned neural networks'. This paper has been accepted for conference L4DC2021 to be published as PMLR.
We welcome collab via branching; please provide your identification and contact details in your branch's description!
Our paper is on arxiv here: http://arxiv.org/abs/2012.00138.
If you find information presented in this paper useful, please consider citing our paper:
@misc{li2020robust,
title={Robust error bounds for quantised and pruned neural networks},
author={Jiaqi Li and Ross Drummond and Stephen R. Duncan},
year={2020},
eprint={2012.00138},
archivePrefix={arXiv},
primaryClass={cs.LG}
}
© J. Li, R. Drummond, S.R. Duncan, Department of Engineering Science, University of Oxford.