Anna Korba

Anna Korba is a researcher affiliated with ENSAE & CREST, Institut Polytechnique de Paris, France. She specializes in sampling methods through optimization of discrepancies, focusing on theoretical and applied aspects of optimal transport, Bayesian inference, and gradient flows.

• Research Interests: Machine Learning, Probability Theory, Variational Methods, Non-convex Optimization.
• Recent Work: Interpolating between MMD and χ² divergences, developing mollified interaction energy descent algorithms, and analyzing quantization errors in particle-based optimization.
• Collaborations: Joint work with researchers across institutions including UCL, CMU, and EPFL.

Scientific Contributions: Her publications address challenges in sampling from non-log-concave distributions, fairness constraints in Bayesian neural networks, and geometric interpretations of gradient flows in Wasserstein spaces.