Coinbase upgrades anti-fraud system, integrating machine learning and rule engines, reducing response time to several hours

Mars Finance News, Coinbase states that it is integrating machine learning models with rule engines to optimize the rule creation process in anti-fraud systems, achieving more efficient risk management. It also proposes a dual-track strategy of “models responsible for long-term defense, rules responsible for rapid response,” and builds a unified framework to create a feedback loop between the two: rules are used to detect new types of fraud and are used to retrain models, thereby continuously enhancing overall defense capabilities.
In specific optimizations, Coinbase reconstructs data structures, automates schema evolution, and introduces notebook-based analysis tools, transforming the originally manual rule creation process into data-driven and automatically recommended methods, significantly improving efficiency.
Among them, rule backtesting performance has increased over 10 times, and overall response time has been shortened from several days to a few hours.
Additionally, the new system uses machine learning to recommend parameters, helping to reduce false positives while combating fraud and minimizing impact on normal users.
Coinbase states that the next step will be to promote event-driven automatic rule generation and explore converting high-efficiency rules into model features with a single click, further advancing toward an automated risk management system.