Data Versioning

[ˈdeɪtə ˈvɜrʃənɪŋ]

AI Infrastructure

Last updated: December 9, 2024

Definition

The practice of tracking and managing different versions of datasets used in ML systems.

Detailed Explanation

Manages changes to training data validation sets and test data over time. Includes tracking of data transformations feature engineering steps and dataset lineage. Enables reproducibility and auditability of model training processes.

Use Cases

Dataset management Training data tracking Feature evolution monitoring Data lineage tracking

Definition

Detailed Explanation

Use Cases

Related Terms

Secure Protocols

Canary Deployment

Teleoperation

Help

People also viewed