Advances in Machine Learning at Microsoft Threat Protection presented at BlueHatShanghai 2019

by Christian Seifert,


Summary : A security solution without machine learning is unthinkable today. Decisions need to be made within milliseconds tapping into the signal embedded in petabytes of data touching over a billion users. In this talk, we will share how we apply machine learning at Microsoft Threat Protection. We will talk to the data, platforms, and capabilities that form the foundation of our models followed by deep dives into select ML models. This talk will go beyond describing traditional classification systems, but rather dive into advanced topics, such as building language models, anomaly models, and adversarial defenses.