We are pleased to announce that Anna Korba, Assistant Professor in Statistics at CREST-GENES, Professor at ENSAE Paris and Hi! Paris affiliated member, has been awarded a prestigious ERC Starting Grant for her research project OptInfinite – Efficient infinite-dimensional optimization over measures.

Launched in 2025 for a 5-year duration, the project aims to develop new optimization and sampling methods over probability measures — a central challenge in artificial intelligence and modern machine learning.

Optimization over probability measures has become an increasingly powerful approach to tackling complex problems involving uncertainty. Unlike traditional methods that work with fixed data points, this framework deals with entire distributions, even in very large or infinite-dimensional spaces.

This is especially useful for sampling tasks, which consist in generating representative examples from a distribution or model. Such tasks are crucial in:

• Bayesian machine learning, where sampling helps quantify uncertainty in model predictions;
• Generative modeling, where it is key to generating realistic new data, such as vectorial data (e.g. images).

However, existing methods often fall short: they are computationally expensive, difficult to evaluate, and poorly suited to high-dimensional or complex distributions. Moreover, they are designed to generate vectorial data, but not more complicated structures such as infinite-dimensional data.

With OptInfinite, Anna Korba aims to overcome these challenges by building a unified theoretical and practical framework, drawing on tools from optimal transport and information geometry. The project will:

• develop more efficient and adaptable sampling algorithms;
• design robust evaluation tools to assess the quality of generated samples;
• deliver an open-source software toolkit to make these methods widely accessible.

The approach will be tested on real-world applications, including large-scale AI models, Bayesian inference, biological systems modeling, and more.

These are fundamental problems in statistics and machine learning, that can be useful in various areas where quantifying uncertainty or when accessible data is scarce, such as finance and economics.

🎧 Curious to learn more about Anna Korba’s background and vision? Check out her interview in our CRESTive Minds series:
👉 Episode 3 – Anna Korba