Federated Learning
YunFL platform is based on machine learning and deep learning algorithm and cryptographic protocol security computing framework, to help local data without leaving, training model will be issued to the local server, with the minimum of data interaction to update the model and iteration, on the premise of guarantee the data security, implement effective multilateral joint is calculated.

Product Details

Horizontal Federated Learning

For sample participants with the same characteristics, comprehensive operation is realized, and machine learning modeling is carried out by using isomorphism distributed among all parties, which is suitable for scenarios with similar businesses. The effect of training model can be improved by increasing the number of samples

Vertical Federated Learning

For the data with high sample user overlap and complementary features, the machine learning model is optimized by using gradient descent and other technologies to enrich the sample feature dimensions, which is suitable for cross-industry and cross-institution data union modeling

Federated Transfer Learning

For the data that cannot meet the horizontal and vertical conditions, there is less user and feature overlap. Therefore, transfer learning technology can be adopted to establish a more widely applicable shared machine learning model

Product Features

Rapid deployment

With automatic modeling function, federated learning training and model deployment supporting various machine learning and deep learning, users can easily get started

Visual monitoring

Comprehensive monitoring of training status and training effect, users can see the training process at a glance

Flexible deployment

It supports access to multiple databases, coordination parties and decentralized versions, and flexible private deployment

Safe and reliable

Through federated learning unique algorithm to ensure that data does not go out of the local, shared data minimization, encryption protocol to ensure the security of data interaction

Architecture Diagram

For more information, please contact our sales and technology team
contact us