Robust and adaptive training algorithms.
Training machine learning models amounts to solving stochastic optimization problems at scale. My research focuses on understanding and improving optimization algorithms, particularly with the aim to design robust and easily tunable methods.