LaConic: Learned Compression Pipelines for Time Series Storage and Analytics
This project aims to explore the feasibility and potential of learned lossy compression pipelines with bounded impact on downstream time series analytics. Solving these challenges will benefit a variety of domains, including medical and health data. We aim to contribute to an improved understanding of lossy time series compression as well as instance-optimized representation according to data and operation characteristics and multiple objectives such as compression ratio, compression speed, and impact on the analysis results. To this end, we will devise a principled approach for learned time series compression as well as develop and share a self-contained and extensible open-source toolkit.
Research Group Lead
Kindly take note that only researchers who have received funding from BIFOLD have their individual profiles displayed on www.bifold.berlin.