Category: Highlight

We are proud to announce that Anna Korba, Assistant Professor in Statistics at CREST-GENES, Professor at ENSAE Paris, and Hi! PARIS Affiliate, has been awarded a European Research Council (ERC) Starting Grant for her project OptInfinite.

At this year’s Hi! PARIS Summer School, Anna Korba (ENSAE Paris) took a fresh look at Langevin diffusions, an old idea from physics that’s quietly becoming central to generative modeling. As machine learning and mathematics increasingly overlap, she invites us to pay closer attention to what’s happening under the hood of today’s most talked-about models.

AI is no longer a support tool in biology, it’s becoming a scientific partner.”
At the Hi! PARIS Summer School 2025, Jean-Philippe Vert (Bioptimus) explored how AI is impacti biomedical research. From protein folding breakthroughs like AlphaFold to in silico simulations of disease, Vert made the case for a new era where biology’s complexity meets AI’s learning power. The next frontier? Models that understand life across molecules, cells, and entire organisms.

Ioannis Stefanou, new Hi! PARIS Chair-holder and Professor at ENSTA, is exploring how artificial intelligence can support the energy transition, starting with the Earth itself. https://youtu.be/vuz7vqlf-QY Can we harness the ground beneath our feet to power a more sustainable future? For Ioannis Stefanou, the answer is yes, and the key lies in combining deep engineering expertise with artificial intelligence. As a Professor at ENSTA and the head of the Geomechanics group at IMSIA, Stefanou is developing new ways to unlock the underground as a source of clean energy and storage. Now, with his appointment as a Hi! PARIS Chair-holder, he’s taking this work even further, bringing together AI, mechanics, and physics to build smarter and more sustainable energy systems. From geothermal to hydrogen: exploring the potential below Stefanou’s research focuses on technologies like deep geothermal energy, CO₂ and hydrogen storage, all of which could play a central role in tomorrow’s energy landscape. But their development requires precise scientific understanding and careful design. “AI helps us explore and control complex underground systems,” he explains. “But we don’t rely on AI alone. We guide it with physical laws and human expertise.” — Ioannis Stefanou, Hi! PARIS Chair Holder This approach, known as physics-informed AI, ensures that machine learning models remain grounded in real-world behavior. It’s a way to build more transparent, explainable, and trustworthy systems that complement, rather than replace, scientific knowledge. Where disciplines meet Stefanou’s work sits at the intersection of mechanics, control theory, and AI, a space he sees as full of opportunity for innovation. “Some of the most interesting ideas emerge when disciplines come together,” he says. “Hi! PARIS offers a dynamic environment where science, engineering, and data can truly interact. It’s an ideal place to collaborate across fields and move research into action.” His own path reflects this interdisciplinarity: a PhD from the National Technical University of Athens, research leadership roles in Europe, and two ERC grants, including one currently underway. A catalyst for sustainable innovation At its core, Stefanou’s work supports the broader goal of sustainable development, creating new tools and models that help scientists and engineers make informed decisions about underground energy systems. “AI is accelerating science in powerful ways,” he says. “Used carefully, it can help us better understand the Earth, design smarter systems, and contribute to a more sustainable energy future.” With his appointment at Hi! PARIS, Ioannis Stefanou joins a growing community of researchers dedicated to using data and AI to serve society, from the ground up. “It’s not about replacing what we know, it’s about expanding what we can do, together.” — Ioannis Stefanou, Hi! PARIS Chair Holder

Luiz Chamon joins Hi! PARIS as a new Chair-holder, bringing a fresh perspective on the mathematical foundations of AI, and what it takes to make machine learning trustworthy by design. https://www.youtube.com/watch?v=0JdLWC_h66o&list=PLneUHBsFcQY71YRntFUdNfEGqzkPCM7n6&index=12 Luiz Chamon’s career has taken him from São Paulo to Philadelphia, from Berkeley to Stuttgart, and now to École polytechnique and Hi! PARIS, where he’s at the frontier of AI and engineering. But his question has remained the same: how can we design intelligent systems that truly serve human needs? “I’m interested in how systems learn from data — but also in how they do it,” Chamon explains. “We need learning methods that don’t just optimize accuracy, but meet strict requirements: fairness, robustness, consistency with science.” That goal calls for a deeper rethink of the foundations of AI. “Too often, constraints like safety or fairness are treated as afterthoughts. I believe they should be built in from the start.” — Luiz Chamon, Hi! PARIS Chair Holder Engineering intelligence, not just optimizing it Chamon’s work looks under the hood of machine learning, using tools from control theory, optimization, and signal processing. His ambition is to shift the focus away from trial-and-error learning toward a model where requirements guide the design of intelligent systems, what he calls “requirement-driven learning.” “It’s not about perfection. But it is about knowing what we want our systems to do, and making sure they do it. That’s engineering. That’s design.” He frames this not just as a technical challenge, but as a necessary evolution of the field. “Artificial intelligence today often means discovering patterns in data. But in many cases, we already know what our systems must respect, physical laws, ethical boundaries, domain constraints. Learning should start from there.” The bigger picture: AI as infrastructure At Hi! PARIS, Chamon is joining a community focused on interdisciplinary, responsible AI. His project fits into a broader ambition: to reimagine the role of AI in society, not as an opaque tool, but as a piece of critical infrastructure. “AI is already shaping our world, sometimes in ways we understand, sometimes in ways we don’t,” he says. “To make it sustainable, we need more than technical performance. We need trust, traceability, clarity.” That’s where his research comes in. “Mathematical foundations aren’t just abstract. They’re how we make sure AI works, and works for everyone.” A place to connect For Chamon, joining Hi! PARIS was also a question of context. “What attracted me was the environment. The chance to work across disciplines, connect with partners in science, engineering, business, all in one ecosystem.” He sees this as essential for moving from theory to real-world impact. “The challenges we face, misinformation, bias, instability, aren’t just technical. They’re social, political, economic. And solving them requires teams that reflect that complexity.” With that mindset, Chamon’s arrival marks more than a new Chair. It signals Hi! PARIS’ continued investment in building AI not just as a technology, but as a shared responsibility. “I want to help shift the mindset, from artificial intelligence as something that ‘emerges’ from data to something we build, together, with intention.”

How is AI changing the way we think about financial markets? In this talk, Charles-Albert Lehalle draws on experience across academia and industry to explore what’s really shifting, from the use of alternative data to the role of retail investors.

From July 13 to 19, Hi! PARIS researchers will take part in the International Conference on Machine Learning (ICML 2025), one of the world’s leading conferences in artificial intelligence, held this year in Vancouver, Canada. A total of 27 research papers from Hi! PARIS affiliated teams have been accepted following ICML’s highly selective peer-review process, recognizing the excellence and innovation of our researchers across partner institutions. This strong presence reflects Hi! PARIS’s scientific leadership and deep engagement in frontier AI research. The accepted papers span a wide array of domains, including machine learning theory, optimization, generative models, privacy, and multi-agent systems, and demonstrate the center’s interdisciplinary strength. By contributing to ICML 2025’s core sessions and workshops, Hi! PARIS reaffirms its commitment to advancing AI research for science, business, and society at the highest international level. Congratulations to our researchers! Here is a list of papers accepted at ICML 2025 that include at least one author affiliated with Hi! PARIS: Title Hi! PARIS Authors All Authors Branches: Efficiently Seeking Optimal Sparse Decision Trees via AO* Albert Bifet, Jesse Read Ayman Chaouki, Jesse Read, Albert Bifet To Each Metric Its Decoding Thomas Bonald, Matthieu Labeau Roman Plaud, Alexandre Perez-Lebel, Matthieu Labeau, Antoine Saillenfest, Thomas Bonald Scaling Laws for Forgetting Marco Cuturi Louis Béthune, David Grangier, Dan Busbridge, Eleonora Gualdoni, Marco Cuturi, Pierre Ablin Shielded Diffusion Marco Cuturi Michael Kirchhof, James Thornton, Louis Béthune, Pierre Ablin, Eugene Ndiaye, Marco Cuturi Misspecification in Simulation-based Inference Marco Cuturi Antoine Wehenkel, Juan L. Gamella, Ozan Sener, Jens Behrmann, Guillermo Sapiro, Jörn Jacobsen, Marco Cuturi TabICL Gael Varoquaux Jingang QU, David Holzmüller, Gael Varoquaux, Marine Le Morvan Byzantine Robust Gossip Aymeric Dieuleveut Renaud Gaucher, Aymeric Dieuleveut, Hadrien Hendrikx Scaffold with Stochastic Gradients Aymeric Dieuleveut, Alain Oliviero Durmus, Eric Moulines Paul Mangold, Alain Oliviero Durmus, Aymeric Dieuleveut, Eric Moulines Compressed and Distributed Least-Squares Aymeric Dieuleveut Constantin Philippenko, Aymeric Dieuleveut Discrete Markov Probabilistic Models Alain Oliviero Durmus Le Tuyet Nhi PHAM, Dario Shariatian, Antonio Ocello, Giovanni Conforti, Alain Oliviero Durmus Prediction-Aware Learning Alain Oliviero Durmus, Eric Moulines Aymeric Capitaine, Etienne Boursier, Eric Moulines, Michael Jordan, Alain Oliviero Durmus Mixture-Based Framework for Diffusion Models Alain Oliviero Durmus, Eric Moulines Yazid Janati el idrissi, Badr MOUFAD, Mehdi Qassime, Alain Oliviero Durmus, Eric Moulines, Jimmy Olsson Differential Privacy for MCMC Alain Oliviero Durmus Andrea Bertazzi, Tim Johnston, Gareth Roberts, Alain Oliviero Durmus Asymmetric Actor-Critic Algorithms Damien Ernst Gaspard Lambrechts, Damien Ernst, Aditya Mahajan Score-Based Generative Models in W2 Marta Gentiloni Silveri Marta Gentiloni Silveri, Antonio Ocello Sliced-Wasserstein Distance Analysis Anna Korba Christophe Vauthier, Anna Korba, Quentin Mérigot Density Ratio Estimation Anna Korba Hanlin Yu, Arto Klami, Aapo Hyvarinen, Anna Korba, Lemir Omar Chehab Wasserstein Gradient Flows Anna Korba Clément Bonet, Christophe Vauthier, Anna Korba Learning of Continuous Markov Semigroups Karim Lounici Vladimir Kostic, Karim Lounici, Hélène Halconruy, Timothée Devergne, Pietro Novelli, Massimiliano Pontil GNN with GMM Augmentation Johannes Lutzeyer, Michalis Vazirgiannis Yassine Abbahaddou, Fragkiskos Malliaros, Johannes Lutzeyer, Amine Aboussalah, Michalis Vazirgiannis TRPO in Mean Field Games Eric Moulines Antonio Ocello, Daniil Tiapkin, Lorenzo Mancini, Mathieu Lauriere, Eric Moulines Conditional Coverage with Conformity Scores Eric Moulines Vincent Plassier, Alexander Fishkov, Victor Dheur, Mohsen Guizani, Souhaib Ben Taieb, Maxim Panov, Eric Moulines Efficient On-Device Learning Van-tam Nguyen, Enzo Tartaglione Le-Trung Nguyen, Aël Quélennec, Van-Tam Nguyen, Enzo Tartaglione Pareto-Optimality in One-Max-Search Vianney Perchet Ziyad Benomar, Lorenzo Croissant, Vianney Perchet, Spyros Angelopoulos Last Iterate Convergence for Uncoupled Learning Vianney Perchet Côme Fiegel, Pierre Menard, Tadashi Kozuno, Michal Valko, Vianney Perchet Quantifying Treatment Effects Erwan Scornet Ahmed Boughdiri, Julie Josse, Erwan Scornet Prediction via Shapley Value Regression Michalis Vazirgiannis Amr Alkhatib, Roman Bresson, Henrik Boström, Michalis Vazirgiannis