MLSEC-06: Enforce data lineage
Monitor and track data origins and transformations over time. Strictly control data access. Perform preventative controls, auditing, and monitoring to demonstrate data lineage. Implement integrity checks against training data to detect any unexpected deviances caused by loss, corruption, or manipulation. Data lineage enables visibility and helps tracing root cause of data processing errors.
Implementation plan
-
Track records for any update - Create and store information about the steps of a ML workflow from data preparation to model deployment using Amazon SageMaker AI ML Lineage Tracker. With the tracking information you can:
-
Reproduce the workflow steps, track model and dataset lineage, and establish model governance and audit standards.
-
Consider origin data to be the source of truth.
-
Ingest and process derived datasets and retain mappings throughout the process. Iterate from the end result back to the original data element.
-
Apply these concepts not just to data, but also the code, models, pipelines, and infrastructure. Validate that you can trace and audit any activity against data, pipeline actions, or machine learning models.
-
Documents
Blogs
Examples
