Paper on data management for Machine Learning accepted to SIGMOD’25.

The paper "Modyn: Data-Centric Machine Learning Pipeline Orchestration" by Maximilian Böther, Ties Robroek, Viktor Gsteiger, Robin Holzinger, Xianzhe Ma, Pınar Tözün, and Ana Klimovic was accepted to SIGMOD'25 in Berlin.

The paper introduces Modyn, a data-centric ML pipeline orchestrator, which executes continuously running ML pipelines using data selection and triggering policies. It also provides a theoretical framework on how to evaluate ML pipelines. The system can be found at external page https://github.com/eth-easl/modyn.