Dr. Jianzong Wang

Deputy Chief Engineer, Ping An Technology

Supernova Award Category: 

Data-Driven Digital Networks (DDNs) and Business Models

The Organization: 

Ping An Technology’s Federated Learning Team is devoted to implement federated intelligence, a new technology that develops machine learning models for the sensitive industry without data aggregation, to resolve the “Isolated Data Island” problem and leverage the value of data for AI application. Our goal is to build a well-rounded federated intelligence ecosystem across industries and lead the trend of privacy-preserving data utilization in business application.

The Problem: 

High quality and a high volume of data has become crucial for fin-tech businesses to build their core competitiveness in Artificial Intelligence (AI). Considering the data privacy and security, the “Isolated Data Island” problem troubles Ping An Group, where 30 subsidiaries focus mining big data values for fin-tech applications under regulations and laws. Without a secure, trusted multi-source data collaboration solution, it would be difficult to break these silos of multi-source data.

The Solution: 

Under Dr. Wang’s leadership, federated learning team uses federated intelligence combined with Intel(R) Software Guard Extensions (Intel(R) SGX), a trusted execution environment for computation built on secure-oriented hardware, to provide reliable security protection for data owners to build and conduct research on AI models without the risk of data leakage.

The Results: 

With federated intelligence, we empower over 2k data-driven scenarios inside Ping An Group. More than 2.2k data scientists have built over 15k federated models.

Federated intelligence has become the core technology in Ping An. It helps with tremendous improvement in like risk management, product recommendation and precision marketing. More than 12 sub-companies in Ping An Group can leverage data value without leaking the raw data to train federated models like risk management models.


Before federated intelligence, data collaboration is difficult for both cross-industry and interagency.

After the federated intelligence project, no raw data exchanged or transferred during the entire data modeling process, avoiding the risk of data leakage. Realizing data’s value for corporations with federated intelligence. The following are outcomes of the use of federated data: 

  • ~15k federated models in fin-tech area are trained by 2.2k data scientists
  • Protect ~9PB secured data while taking advantages of the data for better model performance, including 18% accuracy improvement for anti-fraud detection model and increasing 30% accuracy for anti-money laundering
  • Save about $ 280 million risk costs from fraud federated model.
  • 5% revenue improvement by precision marketing
The Technology: 

Federated Learning Team is committed to develop federated intelligence includes the most commonly used machine learning algorithms like logistic regression, gradient based decision tree are provided, as well as deep learning algorithms like convolutional neural networks, recurrent neural networks, and long-short time memory. It also includes secure channel communication, federated feature engineering, federated incentive mechanism, homomorphic encryption, and secure multiparty computation.

Disruptive Factor: 

Dr. Wang is leading and participating in building an industry standard of federated learning worldwide. The federated incentive mechanism encourages all the organizations to contribute the data asset into federated intelligence ecosystem.

Shining Moment: 

Ping An Technology launches federated intelligence system at the beginning of 2020. The system collaborates with more businesses and institutions to break down data barriers and promote the rapid development and application of federated intelligence in all walks of life.

About Ping An Technology

Ping An Technology is the core technology arm of Ping An Group, using AI, cloud and other cutting-edge technologies to develop and operate mission-critical platforms and services that support financial services, medical health, automotive services, real estate services, and smart cities.