Enabling Data Governance with a Feature Store
Data has become a regulated market with regulations, such as GDPR and CCPA, defining standards for how organizations secure their customers’ data. Organizations need full visibility and control of how all data assets are produced and used – including the data used to train and serve models for AI-enabled products. This problem can be even more pertinent in industries with even tighter regulation, such as healthcare and finance.
A Feature Store can provide an Enterprise with data governance and security for AI. The Feature Store can exercise authority and control (access control, monitoring, auditing, and provenance) over all data assets used in machine learning. Adding a feature store to machine learning pipelines can help ensure compliance with regulatory requirements, making it easier to debug pipelines, automate the operation of pipelines, inspect historical runs of pipelines and their output artifacts. It also ensures the integrity of pipelines and their artifacts.
The Hopsworks Feature Store provides unique support for access control, custom metadata, free-text search, and provenance for features and training datasets.
Watch this video to learn more about best practices for securing and governing your data assets for machine learning. We explore end-to-end model lineage from pre-processing steps to its deployment and maintenance in production. We discuss and go through an interactive demonstration of how we can solve security and governance problems in the Hopsworks Feature Store.
centos 7
