Ilnura Usmanova, Postdoc 

Research interests

My work is mostly focused on developing safe learning and optimization algorithms for convex and nonconvex constrained problems. Particular optimization domains of interest include robust optimization, chance constraints, derivative-free optimization algorithms, large-scale optimization, online learning. In future, I consider developing provably safe feedback control policies with application to robotics in uncertain environment, personalized medicine, etc.

Selected Publications

• Safe Convex Learning under Uncertain Constraints.
  Ilnura Usmanova, Andreas Krause, and Maryam Kamgarpour
  Proceedings of the 22nd AISTATS, PMLR: Vol. 89, 2019

• Safe non-smooth black-box optimization with application to policy search.
  Ilnura Usmanova, Andreas Krause, and Maryam Kamgarpour
  Learning for Dynamics and Control, 2020

• Fast Projection Onto Convex Smooth Constraints
  Ilnura Usmanova, Maryam Kamgarpour, Andreas Krause, Kfir Levy
  Proceedings of the 38th International Conference on Machine Learning, 2021.

• Log Barriers for Safe Black-box Optimization with Application to Safe Reinforcement Learning
  Ilnura Usmanova, Yarden As, Maryam Kamgarpour*, Andreas Krause*
  Under review