Evolving our machine learning to stop mobile bots

This post details the evolution of Cloudflare's machine learning models for bot detection, with a specific focus on improving performance for mobile application traffic. It outlines the process of data gathering, feature selection, model building (using Catboost), offline and online monitoring, and deployment strategies. The key contribution is the identification and incorporation of new datasets representing legitimate mobile app traffic, which were previously under-represented, leading to significant reductions in false positive rates for mobile applications.