AI for Global Goals | OxML

💙 That's a wrap on MLx Representation Learning & Generative AI at OxML Oxford Machine Learning Summer School 2026! 💙 What an incredible four days of learning, collaboration and new friendships. Thank you to everyone who joined us and made this track so special - your curiosity, enthusiasm and passion for research are what make the OxML community so inspiring. A special congratulations to everyone who took part in the Poster Presentation competition. Thank you for sharing your exciting research and contributing to so many great conversations throughout the week! To everyone who helped make this programme possible, and to every attendee who brought such fantastic energy - thank you! We hope to see many familiar faces again next year! #OxML2026 #MachineLearning #RepresentationLearning #GenerativeAI #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

🤖 Day 4 of MLx Representation Learning & Generative AI at OxML | Oxford Machine Learning School 2026 with Kevin Buzzard! 🤖 📣 Our speaker was Kevin Buzzard, Professor of Pure Mathematics at @imperialcollege and maintainer of Mathlib, with a talk on "Why formalize mathematics?"! Kevin's core point was that mathematics has always split into computation (a science, with a method) and reasoning (still an art) — and that formalisation, using tools like the Lean theorem prover, is finally turning proof itself into something machine-checkable. He traced this from the slow, manual decade of building mathlib by hand, through the Polynomial Freiman-Ruzsa formalisation in 2023 that took 20 humans a month, to recent AI systems autoformalising over a million lines of Lean in weeks. His take-home: as AI starts generating mathematics faster than any human can read or trust it, formalisation stops being a nice-to-have and becomes the only way to actually know whether a machine's proof is real - checking, not producing, is where the value shifts. 👏 Thank you, Kevin! #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

📊 Day 4 of MLx Representation Learning & Generative AI at OxML | Oxford Machine Learning School 2026 with @petergostev! 📊 📣 Our first speaker was Peter Gostev, AI Capability Lead at Arena, with a talk on "Benchmarking @ Arena"! Peter's core argument was that no single number captures what a model can actually do — every benchmark only sees one slice of a much bigger, messier shape. He took us through how that plays out in practice: fixed test sets like GPQA and SWE-bench that get saturated and gamed, blind head-to-head comparisons on Arena that reveal real user preference instead, and newer agentic evaluations that track whether a model can actually get real work done, recover from its own mistakes, and take direction. The talk built toward a simple takeaway: we're good at measuring narrow task performance, but still bad at measuring the things that matter most for real usage — whether a model can discover something genuinely new (research), whether it holds up consistently day after day (reliability), and whether it makes sound calls in messy, real-world situations (judgment). Benchmarking, in his view, is racing to catch up with what these models are actually being asked to do. 👏 Thank you, Peter! @GRESEARCHjobs #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

👏 Huge congratulations to our poster presentation winners of MLx Representation Learning & Generative AI at OxML | Oxford Machine Learning School!👏 🏆 Xuerui Zhang — "Heterogeneity-Aware Distillation for Lifelong Heterogeneous Learning" 🏆 Opeyemi Osakuade — "What Speech Tokenisers Forget: Evaluating Tone Retention in Discrete Speech Units" Very well deserved! Thank you for sharing your incredible and inspiring research! A huge thank you to Charles Martínez PhD from @GRESEARCHjobs for presenting the prizes 👏 ✨ #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

✨ The wonderful @balcipelin joined us at OxML Summer School for her second time! It has been such a pleasure welcoming you back to Oxford and seeing you once again as part of our growing community, well done on the winning shots!✨ Thank you for bringing your enthusiasm, curiosity and positive energy throughout the week. We hope you had another unforgettable experience, and we'd love to welcome you back again next year! 💙 #OxML2026 #MachineLearning #OxfordMachineLearning #AIforGlobalGoals #OxfordAI

Over the past two days, we hosted our attendees for a poster presentation competition as part of OxML | Oxford Machine Learning School 2026 and the floor was now theirs 💪 Across two days, 12 researchers from across the world presented their work, fielded questions, and pushed each other's thinking forward. The quality and range of ideas on display was a genuine highlight of this year's programme. Congratulations to everyone who presented, and thank you for bringing your research to OxML! 🎉🎉 @GRESEARCHjobs #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

🎉 OxML | Oxford Machine Learning School 2026 has been an incredible experience, and it simply wouldn't have come together without the four brilliant people who chaired this year's programme! - Mona Alinejad, D.Phil. (Oxon) @monalinejad, our General Chair, Founder and CEO of @elandiai and Founder of @GlobalGoalsAI, brought vision and energy to everything from day one, and set the tone for a programme built on ambition and purpose✨ - Reza Khorshidi, D.Phil. (Oxon) @rezakhorshidi, our Programme Chair, Expert Partner at Bain & Company and Visiting Fellow at @UniofOxford, shaped a programme that balanced academic rigour with real-world relevance, and his experience and generosity made all the difference 💫 - Yali Du @yalidux, our Area Chair, Associate Professor in AI at @KingsCollegeLon and Turing Fellow at The Alan Turing Institute, helped curated a line-up of speakers that pushed us to think differently about AI, and her expertise shaped this track from the ground up 👏 - Sahar Vahdati @SaVahdati, our Poster Chair, Professor of AI for Science at Leibniz Universität Hannover, gave so many researchers a platform to share their work and be seen, and her care for the community showed in every detail 💪 💕 Three of our four chairs this year are women leading at the very top of AI research and industry, and that is worth celebrating. Representation like this doesn't happen by accident, and we are grateful for the doors it opens for everyone who attended. Thank you all, endlessly, for the work, the care, and the leadership you brought to this year's programme. OxML would not be what it is without you. ❤️ #OxML #artificialintelligence #summerschool #oxml26

@balcipelin Thank you, Pelin!🙌🙌

The third day of #OxML26 is complete! Great lectures, inspiring discussions, and fantastic people. It’s wonderful to be back in Oxford! 😊

🎉 Day 4 | Final Day of MLx Representation Learning & GenAI 2026 It's the final day of MLx Representation Learning & GenAI 2026! Thank you for being part of an incredible week of learning, collaboration, and inspiring discussions. Today, we conclude the programme with talks from: 🧠 Peter Gostev (Arena AI) Benchmarking @ Arena 📐 Kevin Buzzard(Imperial College London ) Why Formalize Mathematics 🛡️ Fazl Barez (University of Oxford) AI Safety and Alignment If you're joining us today, make the most of the final sessions, ask questions, connect with fellow participants, and don't forget to share your highlights using #OxML2026. @petergostev @imperialcollege @arena @UniofOxford @FazlBarez

📊 Day 3 of MLx Representation Learning & Generative AI at OxML | Oxford Machine Learning School 2026 with Alexander Tong! 📊 📣 Our second speaker was Alex Tong, from AITHYRA Research Institute for Biomedical Artificial Intelligence with a talk on "Continuous-Time Generative Modeling: Flow Matching and Optimal Transport"! Alex explored how flow matching, the technique powering today's leading image and video generators, learns to turn random noise into realistic data by framing generation as a dynamical system rather than a black box. He unpacked why these models beat older approaches like GANs on stability and flexibility, how optimal transport can straighten the paths for faster sampling, and grounded it all in his own work applying flow matching to single-cell biology to reconstruct how cells evolve over time. A great dive into the maths and intuition behind modern generative AI! 👏Thank you, Alex! @GRESEARCHjobs #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

📊 Day 3 of MLx Representation Learning & Generative AI at OxML | Oxford Machine Learning School 2026 with Yali Du! 📊✨ 📣 Kicking off Day 3 was Yali Du, Associate Professor at @KingsCollegeLon as well as our Area Chair, with a talk on "Human-Centric Cooperative AI: Evaluating, Aligning, and Governing Multi-Agent Systems"! Her talk covered three themes in moving from individual agents to cooperative agent societies: evaluating how agent populations behave under mixed-motive settings (SocialJax, ICLR '26; LLM population studies, AAMAS '25), aligning multi-agent behaviour using imperfect human feedback (M3HF, ICML '25), and constraining agents with enforceable safety rules derived from natural language (SMALL, AAAI '26). She closed by arguing that the future of agentic AI lies not just in autonomous agents, but in governable agent societies. 👏Thank you so much Yali! @GRESEARCHjobs #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

🏆 The trophies are ready for tomorrow's poster presentation prizes at OxML, presented by G-Research! 🏆 A huge thank you to @GRESEARCHjobs for sponsoring us and to the incredible Dr Charles Martínez PhD for presenting the award for best poster tomorrow. Catch the G-Research team at their booth today! 💪 #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

📊 Day 2 of MLx Representation Learning & Generative AI at OxML | Oxford Machine Learning School 2026 with Paul Liang @pliang279! 📊 📣 Our second speaker of the day was Paul Liang, Assistant Professor at @MIT Media Lab and MIT EECS, and director of the Multisensory Intelligence research group, with a talk on "Multimodal AI." The talk was structured in three parts. First, what multimodal AI actually is: data that is heterogeneous, connected, and interacting across modalities. Second, why it's hard: six core technical challenges - representation, alignment, reasoning, generation, transference, and quantification. Third, what's next: multimodal foundation models and how LLMs are adapted for multimodal generation, followed by his group's own work extending AI into new modalities like touch (OpenTouch, tactile sensing gloves) and smell (SmellNet, AromaGen), and finally self-evolving multimodal agents that manage their own memory and reasoning (MEM1, Propose-Solve-Verify, and CORAL for multi-agent evolution). 👏Thank you, Paul! @GRESEARCHjobs #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

📊 Day 2 of MLx Representation Learning & Generative AI at OxML | Oxford Machine Learning School 2026 with Ricardo Silva! 📊 📣 Our second speaker of the day was Ricardo Silva, Professor of Statistical Machine Learning and Data Science at @ucl with his talk on "Causal Learning, Discovery and Extrapolation"! Ricardo starts with how correlation between a treatment and an outcome doesn't tell you whether that treatment actually works, and standard machine learning has no way of distinguishing the two. He walked through how causal models formalise this gap using interventions rather than just observations, using Simpson's Paradox to show how the same data can support opposite conclusions depending on what question you're actually asking. From there he moved into causal discovery itself - how to figure out these structures directly from data when running a randomised trial isn't possible - closing with his own ongoing work on making these methods more practical outside idealised assumptions, including early results in psychology and policy experiments. 👏Thank you, Ricardo! @GRESEARCHjobs #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

There really is no more beautiful campus for a summer school than Oxford! #OxML

📊 Day 2 of MLx Representation Learning & Generative AI at OxML | Oxford Machine Learning School 2026 with @tom_rainforth! 📊 📣 Our first speaker was Tom Rainforth, Associate Professor of Statistical Machine Learning (@UniofOxford), with a talk on "Intelligent Data Gathering"! Tom looked at data gathering as a design problem in its own right - the questions you ask shape how much you actually learn. He walked through Bayesian experimental design, measuring information gain as a reduction in entropy and using optimal Wordle strategies as an example, before explaining why the theoretically optimal approach has traditionally been too expensive to run in practice. His own work, Deep Adaptive Design, tackles that: a trained policy network that gathers more information in milliseconds than classical methods manage in hours. He closed with some great applications - LLMs that ask better follow-up questions, diffusion models building police sketches from witness feedback, and AI systems designing their own scientific experiments. 👏Thank you, Tom! @GRESEARCHjobs #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

🔬 Poster session at MLx Rep Learning & Gen AI OxML | Oxford Machine Learning School 2026! 🔬 Today wasn't just about our incredible speakers - it was also a chance for our attendees to take centre stage. Our poster session brought together research from across the cohort, with participants presenting their own work, fielding questions, and connecting with others. Huge thanks to everyone who put together a poster and shared their research with the community - we can't wait to see more over the next few days! #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals

👾 Day 1 of MLx Representation Learning & Generative AI at OxML | Oxford Machine Learning School 2026 with Petar Veličković (@PetarV_93)! 👾 🧬Our second speaker of the day was Petar Veličković, Senior Staff Research Scientist at @GoogleDeepMind and Affiliated Lecturer at @Cambridge_Uni , presenting "Geometric Deep Learning: Grids, Groups, Graphs, Geodesics and Gauges." Petar is one of the co-authors, alongside Michael Bronstein (@mmbronstein), Joan Bruna, and Taco Cohen, of the Geometric Deep Learning book and framework. Their core idea: most real-world learning problems aren't the worst-case, arbitrary functions that make high-dimensional learning so hard in theory — they carry built-in structure inherited from the physical world. Drawing on Felix Klein's 19th-century Erlangen Programme, which defined geometries by their symmetries, the framework shows how architectures as different as CNNs, RNNs, GNNs, and Transformers are really instances of the same underlying geometric idea, and offers a blueprint for building the next ones. 👏 A rigorous session from one of the field's leading voices. Thank you, Petar! @GRESEARCHjobs #OxML2026 #MachineLearning #GenerativeAI #RepresentationLearning #ArtificialIntelligence #OxfordMachineLearning #AIforGlobalGoals