Mohsen Alimomeni

3:00-3:50 PM - TRACK 4, ROOM 170


In this presentation, we will dive into the critical role of network graphs in detecting security threats and explain why they are not merely an option, but an essential tool for modern security systems. We will showcase several real-world examples and describe how network graph solutions have evolved to tackle adversaries of varying complexity, from simple statistical analysis to advanced graph neural networks and deep learning techniques. Specifically, we will focus on the benefits of using network graphs in a semi-supervised manner, showing how they can overcome two significant challenges in threat detection: the lack of complete and accurate labeled data and adversaries that actively attempt to bypass detection through adversarial attacks.

We will also discuss the challenges of deploying network graphs in production, including issues related to cost and scale, and provide insights into effective methods for deploying them at scale. Finally, we will share the results of our experiments with network graphs, demonstrating how they can significantly enhance the accuracy and efficiency of security threat detection. Whether you're a security professional, data scientist, or software engineer, this presentation will provide valuable insights into the power of network graphs and how they can help you stay ahead of the curve in the ever-evolving threat landscape.

Mohsen Alimomeni

Mohsen is currently leading the threat prevention team in Microsoft, building ML solutions to prevent threats. Previously, he led the data science efforts at EA Sports security to protect games against fraud and cheating. He has been working in the intersection of security and machine learning for over 10 years. He also occasionally teaches at universities and colleges mostly in the field of cryptography and security.