November 19, 2020

Time Travel and Provenance for Machine Learning Pipelines

Implicit model for provenance can be used next to a feature store with versioned data to build reproducible and more easily debugged ML pipelines. We provide development tools and visualization support that can help developers more easily navigate and re-run pipelines .


Alexandru A. Ormenisan, Moritz Meister, Fabio Buso, Robin Andersson, Seif Haridi, Jim Dowling
Download Paper


Machine learning pipelines have become the defacto paradigm for productionizing machine learning applications as they clearly abstract the processing steps involved in trans-forming raw data into engineered features that are then used to train models. In this paper, we use a bottom-up method for capturing provenance information regarding the processing steps and artifacts produced in ML pipelines. Our approach is based on replacing traditional intrusive hooks in application code (to capture ML pipeline events) with standardized change-data-capture support in the systems involved in ML pipelines: the distributed file system, feature store, resource manager, and applications themselves. In particular, we lever- age data versioning and time-travel capabilities in our feature store to show how provenance can enable model reproducibil- ity and debugging.
Download Paper