2022 NortonLifeLock Fellowship Winners!

Yue specializes in designing and building scalable anomaly detection algorithms and systems, with realization and applications in rare disease detection, financial fraud detection, intrusion detection, and malware detection. He has published more than 20 papers in leading machine learning venues including JMLR, TKDE, and NeurIPS, and led multiple impactful open-source machine learning initiatives. Among them, his Python Outlier Detection (PyOD) is the most popular anomaly detection system, enabling numerous real-world applications with more than six million downloads and serving as the primary platform for more than three hundred research projects.

In addition to improving the accessibility of anomaly detection, Yue’s work also streamlines the deployment of anomaly detection, which is also important in security applications. In his NeurIPS 2021 paper, he proposes a novel method called MetaOD to automatically select the top detection model based on the underlying task.

Xinlei's research lies in trustworthy machine learning. Concretely, he focuses on the security and privacy issues stemming from new machine learning (ML) paradigms such as graph neural networks and contrastive learning.

His work reveals that, while being powerful, models trained by such ML paradigms are threatened by various security and privacy attacks. His work tries to give a deeper understanding of what makes the ML model vulnerable and how to build a more secure and private version of it.

His research has been published at top conferences/journals such as IEEE S&P, Usenix Security, ACM CCS, AAAI ICWSM, IEEE TMC, and IEEE TDSC.