Rex Lei

Rex Lei is a Researcher in the Algorithms and Complexity Department (D1) at the Max Planck Institute for Informatics, where he conducts theoretical research on replicability for machine learning algorithms since joining in October 2024.

He received his PhD from the University of California San Diego in 2024 under Russell Impagliazzo, with his dissertation "Replicable Learning Algorithms" publicly available online. The introduction is designed for general audiences without mathematical prerequisites.

Lei's research centers on theoretical computer science foundations of machine learning, specializing in replicability, reproducibility, and algorithmic stability. His work bridges learning theory and complexity theory while exploring connections to privacy, adaptive generalization, and robustness against noise like Massart errors. This interdisciplinary approach addresses fundamental reliability challenges in modern learning systems.

His publication trajectory shows increasing integration of replicability with privacy frameworks and generalization theory, demonstrating how stability properties enable reliable machine learning outcomes across diverse theoretical contexts.

Lei actively contributes to the Algorithms and Complexity research group at MPI Informatics, collaborating with leading theorists on foundational problems in computational learning.