/FXRZ

[ICDE'23] A Feature-driven Accurate and Efficient Lossy Compression Framework

Primary LanguageC++Creative Commons Attribution 4.0 InternationalCC-BY-4.0

ICDE'23 FXRZ: A Feature-Driven Fixed-Ratio Lossy Compression Framework for Real-World Scientific Datasets

Authors: Md Hasanur Rahman (mdhasanur-rahman@uiowa.edu), Sheng Di (sdi1@anl.gov), Kai Zhao (kzhao016@ucr.edu), Robert Underwood (runderwood@anl.gov), Guanpeng Li (guanpeng-li@uiowa.edu) and Franck Cappello (cappello@mcs.anl.gov)

This is a feature-driven compressor-agnostic fixed-ratio framework which can efficiently estimate the expected error bound setting based on a user-specified compression ratio.

We compare our framework (FXRZ) with state-of-the art work FRaZ, which is a high-cost generic fixed-ratio framework.