Publications

2024

1. Optimism in the Face of Ambiguity Principle for Multi-Armed Bandits

   Mengmeng Li, Daniel Kuhn, and Bahar Taşkesen

   In Conference on Web and Internet Economics (WINE), 2024

   Abs arXiv

   Unifying Follow-the-Perturbed-Leader and Follow-the-Regularized-Leader algorithms through the optimism in the face of ambiguity principle.

2. A Large Deviations Perspective on Policy Gradient Algorithms

   (α-β) Wouter Jongeneel, Daniel Kuhn, and Mengmeng Li

   In Learning for Dynamics and Control Conference (L4DC), 2024

   Abs HTML

   A unifying perspective on the convergence analysis of policy gradient algorithms beyond parametrization.

2023

1. Policy Gradient Algorithms for Robust MDPs with Non-Rectangular Uncertainty Sets

   Mengmeng Li, Daniel Kuhn, and Tobias Sutter

   Under Revision, 2023

   Abs arXiv

   The first complete solution scheme for robust MDPs with non-rectangular uncertainty sets, which often display statistical optimality properties.

2021

1. Distributionally Robust Optimization with Markovian Data

   Mengmeng Li, Tobias Sutter, and Daniel Kuhn

   In International Conference on Machine Learning (ICML), 2021

   Abs HTML

   An attempt at the tractability of a statistical optimal DRO model.