Elad Hazan
Papers
1
Total Citations
5
H-Index
1
About
Elad Hazan is a leading figure in the theory and practice of optimization, online learning, and control, with a particular focus on bridging the gap between continuous mathematics and machine learning. His foundational work on the online convex optimization framework has profoundly shaped modern algorithmic design, providing the theoretical backbone for adaptive gradient methods widely used in deep learning. Among his most impactful contributions is the development of efficient algorithms for non-smooth optimization and regret minimization, which have garnered thousands of citations and are now standard in the field. Beyond theory, Hazan has pioneered differentiable control through his work on the Deluca library, a natively differentiable physics and robotics environment that enables gradient-based training of control policies. This open-source tool, introduced in a 2021 paper, allows researchers to auto-differentiate through simulation dynamics, dramatically accelerating the development of robotic controllers. His work has earned him numerous accolades, including best paper awards and a prestigious European Research Council grant. For students and researchers, Hazan’s research exemplifies how rigorous theoretical insight can directly enable practical, high-impact engineering solutions.
Research Focus
Key Achievements
Top Papers
- 1