CHINA — From November 14th to 16th, LexisNexis Risk Solutions and a major client conducted privacy computing and federated modeling. This project, under the principle of protecting data privacy, achieved joint modeling, realizing the principle of data being “accessible and invisible."
The "LexisNexis Risk Federated Learning Platform," developed by LexisNexis Risk Solutions, passed the privacy computing product evaluation by the China Academy of Information and Communications Technology in July this year. This evaluation, known as the "Trusted Privacy Computing" assessment, awarded the platform a "Special Evaluation Certification for Basic Capabilities in Federated Learning," signifying that the federated learning platform developed by LexisNexis Risk Solutions has been recognized by professional institutions in aspects such as scheduling management, data processing, algorithm implementation, model effectiveness and performance, and security.
Based on the LNFA platform constructed by LexisNexis Risk Solutions, both parties uploaded their data to this privacy computing platform and carried out processes such as data authorization, sample alignment (PSI), feature selection, feature engineering, model training, and model validation. They tested mainstream federated algorithms such as SecureBoost (XGBoost). The project, compared with traditional centralized modeling, met both parties' expectations in terms of model accuracy and performance.
Senior data scientist Shan Xiang and technical manager Huang Jiyong from LexisNexis Risk Solutions engaged in in-depth discussions with the client on encryption principles, federated learning theories and algorithms, platform architecture, and security frameworks. Shan Xiang emphasized that optimizing model parameters is key to enhancing the performance and stability of federated learning. He believes it's important to meet insurance company clients' needs for visual analysis of model results, ultimately achieving federated learning, analysis, and deployment on the platform.
This collaboration further promoted the implementation of privacy computing platform of LexisNexis Risk Solutions in the insurance industry scenario. Without exposing their respective data and ensuring the privacy and security of both parties, it connected "data islands," realizing the shared interchange of data value and models.
About LexisNexis Risk Solutions
LexisNexis® Risk Solutions harnesses the power of data, sophisticated analytics platforms and technology solutions to provide insights that help businesses across multiple industries and governmental entities reduce risk and improve decisions to benefit people around the globe. Headquartered in metro Atlanta, Georgia, we have offices throughout the world and are part of RELX (LSE: REL/NYSE: RELX), a global provider of information-based analytics and decision tools for professional and business customers. For more information, please visit LexisNexis Risk Solutions and RELX.