Developing a Data-Driven Web Attack Detector

Опубликовано 27 июня 2016, 22:08

Developing a Data-Driven Web Attack Detector This work demonstrates how to benefit from usage logs in the implementation of a web attack detector. A partnership between Bing and Azure has been working on building a web attack detection module in order to have a WAF (Web Attack Firewall) that can block bad requests, with minimal interference in service operation. The path going from traditional regular expression-based approaches to include machine learning modules is not a linear one, and focus on the data has been proven to be more important than experimentation with ML methods in the path towards both accuracy and runtime performance.