DPZ is a lossy compression technique for scientific data based on multi-stage feature extractions, commonly employed in information retrieval. Unlike other exiting techniques where the compression is either done by predicting or bit-plane encoding, DPZ focus on preserving the key data content from each stage to the maximum extent, ultimately elevating the compression ratios. DPZ is written in Python and is an implementation of the overall idea described in the following paper.
Jialing Zhang, Jiaxi Chen, Xiaoyan Zhuo, Aekyeung Moon, Seung Woo Son
"DPZ: Improving Lossy Compression Ratio with Information Retrieval on Scientific Data"
IEEE Cluster, September 2021.
BibTeX citation entry:
@inproceedings {DPZ-Cluster21,
author = {Jialing Zhang and Jiaxi Chen and Xiaoyan Zhuo and Aekyeung Moon and Seung Woo Son},
booktitle = {2021 IEEE International Conference on Cluster Computing (CLUSTER)},
title = {{DPZ: Improving Lossy Compression Ratio with Information Retrieval on Scientific Data}},
year = {2021},
pages = {1-12},
doi = {},
}