Author: Christina Harter
The LF AI & data foundation, which is building an ecosystem to support open source innovation in artificial intelligence (AI), machine learning (ML), deep learning (DL) and open source data projects, today announced feast as its latest incubation project.Feast（Feature StOre is an open source feature store for machine learning.
Today, teams running operational machine learning systems face many technical and organizational challenges:
- The model has no consistent view of feature data and is tightly coupled with the data infrastructure.
- Deploying new features in a production environment is difficult.
- Feature leakage reduces the accuracy of the model.
- Features cannot be reused across projects.
- The operations team cannot monitor the quality of the data provided to the model.
Fast was jointly developed by gojek and Google cloud in 2018 and was open source in early 2019. The plan aims to meet the following challenges:
- Provides a single data access layer that separates the model from the infrastructure used to generate, store, and service characteristic data.
- The creation and use of features are separated through centralized storage, which allows teams to deliver features to production with minimal engineering support.
- It provides real-time and accurate feature data retrieval for model training and online services.
- Encourage reuse of feature parts by allowing organizations to build a shared foundation for feature parts.
- Provide data centric operation monitoring to ensure that the operation team can safely run the production machine learning system on a large scale.
“Feast was created to address the data challenges we face at gojek, while expanding machine learning for car hailing, meal delivery, digital payments, fraud detection, and countless other use cases.” “After the project was open source, we saw an explosion in demand for software, leading to strong adoption and community growth,” said Willem Pienaar, founder of fast. Entering the LF AI & data foundation is an important step towards decentralized governance and wider industry adoption and development. “
“Feast’s entry into the LF AI & data foundation is both an important milestone in the project and recognition of the project’s progress in solving some of the most difficult problems in machine learning data production,” said Jeremy Lewi, founder of kubeflow. Technologies like coast have the potential to shape the machine learning stack of the future, and with its incubation at lf AI & data, the project now has the ideal environment to expand its community and build a first-class open source feature store. “
Dr Ibrahim Haddad, executive director of LF AI & data, said: “we are very excited to welcome fast to LF AI & data and help it flourish in a supplier neutral environment under an open governance model. With the addition of feast, we will increase the number of managed projects under the data category, and expect our data projects to strengthen cooperation with all other projects to promote innovation in open source technologies for data, analysis and artificial intelligence. “
LF AI & data supports projects through a wide range of services, and the first step is to join as an incubation project. LF AI & data will support feast’s neutral and open governance to help foster project growth. Please checkfile, start using feast today. stayGitHubTo learn more about feast, be sure to joinFEAST-AnnounceandFEAST-Technical-DiscussMailing list, join the community and keep up to date.
Welcome to fast! We look forward to the continued growth and success of this project as part of the LF AI & data foundation. To learn how to host open source projects with us, visitLF AI & data website。
Linux foundation is a non-profit organization and an important part of technology ecosystem.
The Linux foundation supports the creation of a sustainable open source ecosystem by providing financial and intellectual resources, infrastructure, services, activities, and training. In the creation of shared technology, the Linux foundation and its projects have formed an extraordinary successful investment through joint efforts. Scanning two-dimensional code concerns LFAPAC WeChat official account.