Uber introduced Michelangelo Palette, the first Feature Store in 2017. Feature Store enables the discovery, documentation, and reuse of features. It is a feature computation and storage service that enables features to be registered, discovered, and used for ML pipelines and by online applications for model inferencing. Feature Stores stockpile swathes of feature data and offer low latency access to features for online applications. It also ensures consistent feature computations across the batch and serving APIs. We have compiled a list of the top Feature Stores of leading companies.
Amazon SageMaker Feature Store
SageMaker Feature Store is a fully managed, purpose-built repository to store, update, retrieve, and share machine learning features.It can ingest data from multiple sources and tags and indexes features to helps users easily discover them through a visual interface in SageMaker Studio.
THE BELAMY
Sign up for your weekly dose of what's up in emerging technology.
Image: Amazon
It also provides a unified store for features during training and real-time inference and doesn’t require any extra code or manual maintainence. It tracks metadata of stored features to help user query the features for the right attributes in batches or in real time using Amazon Athena. The Feature Store easily integrates with Amazon SageMaker Pipelines to create, add feature search and discovery, and reuse automated machine learning workflows.
Vertex AI Feature Store
The Vertex AI Feature Store from Google allows data scientists to focus on the feature computation logic rather than on the problems that come with deploying features in the production stage. Essentially, it is a fully managed centralised repository for organising, storing, and serving ML features. Vertex AI Feature Store provides search and filter capabilities for users to easily discover and reuse existing features. It also helps detect significant changes to your feature data distribution over time ( called drift). Additionally, it constantly tracks the distribution of feature values ingested into the Feature Store.
Feathr-LinkedIn
LinkedIn defines Feathr as an abstraction layer that offers a common feature namespace for defining features and a common platform for computing, serving, and accessing them ‘by name’ within ML workflows.
Feathr enhances machine learning productivity by reducing the need for individual teams to manage feature pipelines with the option of sharing features easily across projects. Feathr also has advanced support for feature transformations, enabling users toexperiment with new features based on raw data sets. Just a few days back, LinkedIn open sourced Feathr.
Hopsworks
The Hopswork Feature Store allows scaling a model to hundred times. It improves the quality and re-usability of features and models to develop solutions for business use cases. The open-source Feature Store works for environments such as Amazon, Google and Azure.
ML Lakes-Salesforce
ML Lake allows Salesforce application developers and data scientists to easily build machine learning capabilities on customer and non-customer data. It centralises the security controls required for maintaining trust. It assists developers in managing data pipelines, storage, security and compliance.
Image: Salesforce
Eli Levine from Salesforce engineering said ML Lake is deployed in multiple AWS regions as a shared service for use by internal Salesforce teams and in applications running in both public cloud providers and Salesforce’s own data centres.
Overton-Apple
Overton also wants to help developers to concentrate on higher-level tasks instead of lower-level machine learning tasks. Overton supports developers and engineers in building, monitoring, and improving production machine learning systems. Apple adds that Overton “automates the life cycle of model construction, deployment, and monitoring by providing a set of novel high-level, declarative abstractions.”
