Ant Group’s Financial AutoML Platform Wins CCF Technology Advancement Excellence Award

On October 20, 2020, the China Computer Federation (CCF) announced the 2020 CCF Science and Technology Award results, and Ant Group’s financial intelligent AutoML system won the CCF Technology Advancement Excellence Award, the highest honor CCF bestows for technology application, indicating industry recognition of Ant’s innovation in financial AI.

Building financial intelligence by enhancing automated modeling capabilities in financial scenarios is crucial for applying AI technologies in the finance industry. The main challenge is the growing demand for intelligent solutions versus a severe shortage of AI talent, and the solution lies in breaking through the technical barriers centered on automated machine learning.

The Ant Group financial intelligent AutoML project team, in collaboration with Zhejiang University, began technical research in early 2017 and was the first to develop an industrial‑grade financial AutoML platform (FinAutoML) with fully independent intellectual property, achieving large‑scale application. The main innovations are:

Construction of an industrial‑grade full‑link automated modeling algorithm library. Designed algorithms for automated feature engineering, hyper‑parameter optimization, neural architecture search, model compression, and model interpretability. The proposed SAFE algorithm efficiently derives effective features, increasing risk‑control scenario coverage by 18%.

Proposal of a high‑performance AutoML architecture. Compared with commercial and open‑source systems, it significantly improves efficiency in feature automation, model compression, and deep neural network architecture search.

Innovation of an automated machine modeling system supporting diverse financial data. It offers high automation, scalability, and usability, covering graph, privacy‑preserving, and transfer learning algorithms, and provides visual programming capabilities that lower the barrier to AI application.

The project was the first to achieve large‑scale automated intelligent modeling in internet finance, supporting Ant Group’s core payment, insurance, and risk‑control services. Compared with commonly used industry methods, it improved financial AI modeling efficiency by 50% and risk discrimination by 20%, delivering higher‑quality services to users and merchants. The project has generated over 30 authorized invention patents and published more than 40 papers in major AI conferences and journals such as KDD and VLDB.

Senior Ant Group technical expert Zhou Jun stated, “Automated machine learning provides tremendous support for the development of the AI industry, helping break AI technology barriers across domains and delivering multiple commercial benefits. Ant will continue to explore AutoML deeply, applying AI dividends to more real‑world scenarios for the benefit of users, merchants, and partners.”

— END —