The Linux Foundation Projects
Skip to main content

Discover LF AI & Data Projects with TAC Talks Watch Now ➔

search
LF AI & Data Blog

FATE Release (v.1.10.0), KubeFATE (v.1.10.0) and FedLCM (v.0.2.0) Now Available!

By January 18, 2023No Comments

LF AI & Data is pleased to share that FATE, an LF AI & Data Foundation Incubation-Stage Project, has released new versions of FATE v1.10.0, KubeFATE v1.10.0 and FedLCM v0.2.0 in January 2023. FATE (Federated AI Technology Enabler) is the world’s first industrial-grade federated learning open source framework 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 now provides a host of federated learning algorithms, including logistic regression, tree-based algorithms, deep learning and transfer learning. 

Major features and improvements of these releases include:

FATE v1.10.0 release

  • FederatedML
    • Renewed Homo NN: PyTorch-based, support flexible model building:
      • Support user access to complex self-defined PyTorch models or ready-to-use PyTorch models such as DeepFM, ResNet, BERT, Yolo
      • Support various data set types, may build data set based on PyTorch Dataset
      • User-defined training loss
      • User-defined training process: user-defined aggregation algorithm for client and server
      • Provide API for developing Aggregator
  • Upgraded Hetero NN: support flexible model building and various data set types:
    • more flexible pytorch top/bottom model customization; provide access to industry approved PyTorch models
    • User-defined training loss
    • Support various data set types, may build data set based on PyTorch Dataset
  • Semi-Supervised Algorithm Positive Unlabeled Learning
  • Hetero LR & Hetero SecureBoost now supports Intel IPCL
  • Intersection support Multi-host Elliptic-curve-based PSI
  • Intersection may compute Multi-host Secure PSI Cardinality
  • Hetero Feature Optimal Binning now record & show Gini/KS/Chi-Square metrics
  • Host may load Hetero Binning model with WOE score through Model Loader
  • Hetero Feature Binning support binning by user-provided split points
  • Sampler support weighted sampling by instance weight
  • FATE-Flow
    • Add connection test API
    • May configure gRPC message size limit
    • Fix module duplication issue in model 
  • FATE-Board
    • Display SBT leaf node data
    • Support result summary display for Sampler’s new method 
    • Add model summary for new module Positive Unlabeled Learning
    • Improved table display for Binning
    • Data filtering on requested model proto
    • Adjusted Design
    • Improved Logging display adaptation
  • FATE-Client
    • Flow CLI adds min-test options
    • Pipeline adds `data-bind` API, useful for local development 
    • Pipeline may reconfigure role/model_id/model_version, switching `party_id` for prediction task

KubeFATE v1.10.0 release

KubeFATE v1.10.0 supports FATE v1.10.0, FATE-Serving v2.1.6, and brings the following updates:

  • Support dependent distribution in FATE cluster deployed by FATE
  • Support setting password for jupyter notebook
  • KubeFATE service adapts the API of StatefulSet resources
  • Add more pulsar config for external pulsar
  • Support FATE-Flow HA
  • Fix the issue existing claim doesn’t work for nodemanger
  • Fix other bugs

FedLCM v0.2.0 release

The v0.2 release of FedLCM contains a Site Portal service that can be deployed via a standalone, docker-compose approach, decoupling it from the Lifecycle Manager service. In addition to that, Site Portal can now start vertical federated learning jobs. And the Lifecycle Manager service now requires less Kubernetes permissions. Other major features and improvements are:

  • Lifecycle Manager
    • Support deploying and managing KubeFATE v1.9, FATE v1.9 and FATE Exchange v1.9 releases
    • Automatically configure the deployed Site Portal and register it to FML Manager. No manual action is needed
    • Support adding Kubernetes infrastructures using namespace-only permissions
    • Support specifying external engines when deploying FATE clusters, and skip deploying these engines as containers
    • When Lifecycle Manager is running in Kubernetes, the underlying Kubernetes can be added automatically as an infrastructure
  • Site Portal & FML Manager
    • Site Portal and FML Manager can be deployed and running in a standalone fashion using docker-compose. It is still recommended to deploy Site Portal using Lifecycle Manager but it is no longer mandatory
    • When running as a standalone service, Site Portal can work with both FATE with Spark and FATE with Eggroll
    • Support HeteroLR and HeteroSBT type of training and predicting jobs, in addition to the existing HomoLR and HomoSBT jobs
    • Support add existing FATE table into the service for future jobs
    • Support un-registration of the Site Portal from the FML Manager

Any feedback is welcomed so we can further improve the project. To learn more about the FATE v1.10.0, KubeFATE v.1.10.0, and FedLCM v.0.2.0 releases, please review the full release notes. 

Want to get involved with FATE? Be sure to join the FATE-Announce and FATE Technical-Discuss mailing lists to join the community and stay connected on the latest updates.

FATE Key Links

LF AI & Data Resources

Author

Close Menu