Business meeting of staff and management in a printery close to a production line
Projects

LF AI & Data + VMware’s FedLCM Project: Strengthening the Operation of the Federated Learning Platform

VMware recently donated the open source project Federation Lifecycle Manager (FedLCM) to the Linux Foundation’s AI & Data Foundation (LF AI & Data), which aims to streamline the provisioning and management of federated clusters.

FedLCM works with Federated AI Technology Enabler (FATE), a project also hosted by LF AI & Data, to enable enterprises and institutions to collaborate on data while protecting data security and privacy. It implements secure computation protocols based on homomorphic encryption and multi-party computation (MPC). Supporting various federated learning scenarios, FATE provides a host of federated learning algorithms, including logistic regression, tree-based algorithms, deep learning and transfer learning. FATE is one of the largest communities in the federated learning space, with more than 4,000 members from enterprises and research institutes from around the world. The vision of the community is “open source, open community, and co-innovation.” VMware is a board member of FATE’s Technical Steering Committee (TSC).

“We’re thrilled to welcome FedLCM to our community,” said Ibrahim Haddad, Ph.D., General Manager of LF AI & Data. “FATE just joined our portfolio of projects with a strong community presence. The momentum demonstrated by the donation of FedLCM only builds on the progress made by FATE. Both of these initiatives align with our mission of building and supporting an open AI and data community while driving innovations with collaborations with community members.”

“We are very excited to open source and donate FedLCM as an important tool to the federated learning community,” said Henry Zhang, the Chair of FATE’s Development Committee, Engineering Director of Cloud Native Lab, VMware R&D. “It is a long-awaited project, which provides the powerful capabilities to manage the lifecycle of federated learning frameworks, such as FATE and OpenFL. We will continue to collaborate with community members to improve FedLCM and drive its wider adoption.”

An overview of FedLCM

Federated learning (also known as collaborative learning) is a machine learning technique that trains a model across multiple organizations holding local data samples without exchanging them. Orchestrating the federated learning tasks while maintaining data privacy and security is challenging. FedLCM reduces the barriers to using the federated learning platform by providing neutral support to multiple federated learning frameworks. It offers a unified experience to provision and manage different frameworks across organizations in a federation.

Many industries employing AI face challenges, such as data silos, where data is dispersed across isolated sources, compromising data privacy. FedLCM supports multiple infrastructures including Kubernetes. The federated learning frameworks deployed in these infrastructures can collaborate as a federation to break data silos without compromising data privacy.

FedLCM Chart

FedLCM comes with a site management platform called the Site Portal for FATE. The Site Portal is a graphical federated learning task management service that allows users to initiate or join a federal learning task through a web GUI.

FML Site Portal Screenshot

What’s next?

FedLCM will continue to support more management of capabilities of FATE and other open source frameworks like OpenFL. For more information about FedLCM or to contribute to the project, please visit GitHub.

To learn more about other VMware community efforts and contributions to the community of federated learning, check out these blogs:

Pushing the Limits of Network Efficiency for Federated Machine Learning

Cloud-Native Federated Learning and Projects

Stay tuned to the Open Source Blog and follow us on Twitter for more deep dives into the world of open source contributing.