Skip to content

Release 1.0
Compare
Choose a tag to compare
@dylan-fan dylan-fan released this 18 Aug 07:26
· 13734 commits to master since this release
v1.0
09bd5be
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Verified
Learn about vigilant mode.

Major Features and Improvements

This version includes two new products of FATE, FATE-Board, and FATE-Flow respectively, FATE-Board as a visual tool for federation modeling, and FATE-Flow is an end to end pipeline platform for federated learning. This version contains important improvements to the FederatedML, which better tracks the running progress of federated learning algorithms.

FATE-Board

  • Federated Learning Job DashBoard
  • Federated Learning Job Visualisation
  • Federated Learning Job Management
  • Real-time Log Panel

FATE-FLOW

  • DAG defines Pipeline
  • Federated Multi-party asymmetric DSL parser
  • Federated Learning lifecycle management
  • Federated Task collaborative scheduling
  • Tracking for data, metric, model and so on
  • Federated Multi-party model management

FederatedML

  • Update all algorithm modules running mechanism for supporting federated modeling pipeline by FATE-Flow
  • Intermediate statistic result callback is available and visualizable in FATE-Board for all algorithm modules.
  • Support Nesterov Momentum SGD Optimizer
  • Add Homomorphic Encryption Scheme Based on Affine Transforms
  • Support sparse input-format in federated feature binning
  • Update evaluation metrics, such as ks, roc, gain, lift curve and so on
  • Update algorithm's parameter-define class

FATE-Serving

  • Add online federated modeling pipeline DSL parser for online federated inference
Assets 2
Loading