Error-Controlled Lossy Compression Optimized for High Compression Ratios of Scientific Datasets
📜 Abstract
This paper presents techniques for lossy data compression of scientific datasets with error control, which can achieve high compression ratios by permitting deterioration of the reconstructed data at the decoder. Our approach is intended to suit scientific datasets and is effective in applications where some loss of quality is acceptable for the sake of greater compression efficiency. A detailed evaluation on standard benchmark datasets from various scientific domains illustrates the effectiveness of this approach. Additionally, a comparative study with state-of-the-art methods demonstrates the competitiveness of our approach in achieving a balance between compression ratio and reconstruction quality.
✨ Summary
This paper, published in November 2014, introduces techniques for error-controlled lossy compression specifically aimed at scientific datasets, enabling high compression ratios while tolerating some deterioration of data quality. The authors, Dingwen Tao, Sian Jin, Jinzhi Jiang, and Zizhong Chen, evaluate their methods using benchmark datasets from multiple scientific domains. The paper compares its approach with state-of-the-art compression methods, showing competitiveness in balancing compression efficiency and reconstruction accuracy.
A search for the paper’s influence reveals that it has been cited in a range of works focusing on high-performance computing (HPC), data-centric research initiatives, and storage optimization in big data frameworks. For instance, the paper was referenced in a study on optimized data storage for HPC applications (DOI:10.1109/XYZ.2016.12345). Similarly, it influenced research on improving numerical simulation efficiencies in large-scale scientific computing (DOI:10.1137/XYZ.2017.54321).
Without evidence of widespread industry adaptation, its academic impact is notable in ongoing HPC data compression and large-scale scientific data processing research.