Anna Korba
About me
Since September 2020, I am an assistant professor at ENSAE/ CREST in the Statistics Department.
My main line of research is machine learning. I have been working on kernel methods, optimal transport, optimisation, particle systems and preference learning. At the moment I am particularly interested in sampling and optimisation methods.
In 2025, I have been awarded an ERC Starting Grant for my project Optinfinite.
Team
With my academic activities, I am lucky to work closely with the following people:
- Yedidia Agnimo, Cifre PhD, co-advised with Karteek Alahari and Nicolas Chesneau
- Paul Caucheteux, PhD
- Clémentine Chazal, PhD
- Yani Hammache, intern
- Andrea Milone, intern
- Marguerite Petit-Talamon, PhD
- Christophe Vauthier, PhD, co-advised with Quentin Mérigot
- Adrien Vacher, postdoctoral researcher
- Oussama Zekri, PhD, co-advised with Nicolas Boullé
Alumni
News
- July 2026: New preprint, Uniform-in-time Propagation-of-Chaos for Stein Variational Gradient Descent, with Krishnakumar Balasubramanian and Sayan Banerjee. We establish uniform-in-time propagation-of-chaos guarantees for SVGD, including logarithmic rates in broad distributional metrics and parametric rates for finite-dimensional Stein observables.
- June 2026: New preprint, Highly Data Parallelizable Estimation of the Sliced-Wasserstein Distance Using Cumulative Distribution Functions, with Christophe Vauthier and Quentin Mérigot. We introduce sorting-free Sliced-Wasserstein estimators based on projected cumulative distribution functions, enabling massive parallelism and federated computation.
- May 2026: New preprint (accepted as a spotlight at ICML 2026), A Unifying View of Variational Generative Wasserstein Flows, with Paul Caucheteux and Clément Bonet. We develop a unified JKO-based framework for generative Wasserstein flows that connects existing methods and yields new algorithms for f-divergences, integral probability metrics and MMD.
- May 2026: New preprint, Evaluating the Relevance of Uncertainty Estimators for LLM Hallucination, with Yedidia Agnimo, Annabelle Blangero, Nicolas Chesneau and Karteek Alahari. We show empirically that the link between uncertainty estimates and LLM hallucinations is highly variable and sometimes weak across models, benchmarks and hallucination types.
- March 2026: New preprint, Generalized Discrete Diffusion from Snapshots, with Oussama Zekri, Théo Uscidda and Nicolas Boullé. We propose a unified discrete diffusion framework with arbitrary corruption processes and a snapshot-based training objective that improves efficiency and generation quality.
- September 2025: New preprint, A Computable Measure of Suboptimality for Entropy-Regularised Variational Objectives, with Clémentine Chazal, Heishiro Kanagawa, Zheyang Shen and Chris Oates. We introduce computable gradient discrepancies for assessing entropy-regularised variational solutions, generalising the kernel Stein discrepancy beyond Bayesian targets.
- July 2025: New preprint, Kernel Trace Distance: Quantum Statistical Metric between Measures through RKHS Density Operators, with Arturo Castellanos, Pavlo Mozharovskyi and Hicham Janati. We introduce a distributional distance based on Schatten norms of kernel covariance operators that is more discriminative than MMD while retaining favourable sample complexity.
Bio
From December 2018 to August 2020 I was a postdoctoral researcher at Gatsby Unit, University College London (UCL), working with Arthur Gretton.
From October 2015 to October 2018, I was a PhD student at Télécom ParisTech, in the S2A (Signal, Statistics and Learning) team, supervised by Stephan Clémençon .
Before that in 2015, I graduated the Master MVA (Machine Learning and Computer Vision) from ENS Cachan and ENSAE.
More details can be found in my resume [EN].