Privacy Computing Practices at China Construction Bank: From External Data Use to an Enterprise‑Grade Platform
This article details China Construction Bank's journey in privacy computing, covering the history of external data usage, early federated learning experiments, rapid demand growth, the construction of an enterprise‑grade privacy computing platform, its design principles, achievements, and future directions.
The article introduces the Shanghai Big Data Smart Center of China Construction Bank, established in 2015 to provide data analysis, platform technology, and algorithm support for various business lines, and outlines the main topics of the presentation.
External Data Usage History : Defines external data, describes its acquisition methods, and explains how the bank has used external data for marketing, risk control, decision making, and product innovation since before 2015, with a centralized management model introduced in 2017.
Privacy Computing Early Exploration (2020‑2021) : Explains the strategic background of privacy computing, its benefits for data security, minimal data usage, and preventing misuse, and presents a joint modeling case with the Construction Bank Trust Fund using the FATE framework, achieving superior model performance.
Privacy Computing Demand Surge (2022‑2023) : Describes the deployment of a multi‑party computation framework alongside FATE, expanding use cases to secure intersection, anonymous queries, and secure computation, and lists numerous business scenarios with partners such as Meituan, UnionPay, and telecom operators.
Enterprise‑Grade Privacy Computing Platform Construction : Reviews the limitations of the initial platform, outlines the goals for an enterprise‑grade solution that enables data sharing without exposing raw data, and presents design principles focusing on architecture, value creation, regulatory compliance, and cryptographic foundations.
Platform Design and Architecture : Summarizes the four‑layer design (data layer, algorithm layer, service layer, application layer) and the overall architecture diagram that supports joint query, joint computation, and joint modeling across institutions.
Achievements and Future Directions : Highlights awards, patents, and research projects, and outlines future plans to accelerate data‑fusion applications, strengthen ecosystem cooperation, and invest in advanced privacy‑preserving algorithms resistant to malicious and quantum attacks.
DataFunTalk
Dedicated to sharing and discussing big data and AI technology applications, aiming to empower a million data scientists. Regularly hosts live tech talks and curates articles on big data, recommendation/search algorithms, advertising algorithms, NLP, intelligent risk control, autonomous driving, and machine learning/deep learning.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.