معرفی
Adrian Vladu is a permanent researcher at CNRS, affiliated with the Institute of Research in Fundamental Computer Science (IRIF) at Université Paris Cité. He holds a PhD in Mathematics from MIT (2017), advised by Jonathan Kelner and Aleksander Mądry, followed by postdoctoral work at Boston University. His research focuses on optimization (convex/non-convex) and deep learning, emphasizing algorithm design with provable guarantees. Recent work explores refining interior point methods and scalable training techniques for deep models.
Education: PhD in Mathematics (MIT, 2017); Postdoctoral Associate (Boston University). Research interests include combining convex geometry, continuous optimization, and linear algebra to address discrete problems. He also investigates deep learning's theoretical underpinnings, particularly generalization and training efficiency. Current projects aim to develop simple, implementable algorithms with strong theoretical foundations.
Grants: Supported by CNRS and the French National Research Agency. Job Market: Actively seeking tenure-track faculty or industry research roles in optimization, algorithms, or machine learning. Notable contributions include STOC/FOCS/SODA publications on optimization algorithms and ICLR/ICML work on adversarial attacks and distributed training.
