Research Blog
15.01.2026
Blind Matching – Aligning Images and Text Without Training or Labels
MCML Research Insight - With Dominik Schnaus, Nikita Araslanov, and Daniel Cremers
Vision-language models have shown that images and text can live in a shared space: a picture of a “cat” often lands close to the word “cat” in the embedding space. But such models learn this correspondence because they have seen millions of image–caption pairs. What if we had …
08.01.2026
High-Res Images, Less Wait: A Simple Flow for Image Generation
MCML Research Insight - With Johannes Schusterbauer, Pingchuan Ma, Vincent Tao Hu, and Björn Ommer
Image generation models today can create almost anything, like a futuristic city glowing at sunset, a classical painting of your cat, or a realistic spaceship made of glass. But when you ask them to go bigger and sharper, the magic slows down. The process takes longer, eats up more memory, and trying out new versions can feel endless. For quite …
18.12.2025
“See, Don’t Assume”: Revealing and Reducing Gender Bias in AI
MCML Research Insight - With Leander Girrbach, Yiran Huang, Stephan Alaniz and Zeynep Akata
Using AI and LLMs at work feels almost unavoidable today: they make things easier, but they can also go wrong in important ways. One of the trickiest problems? Gender bias. For example, if you ask about someone’s skills from a photo, it may confidently label them a “born leader” or “working well under pressure” with no real …
11.12.2025
From Sitting Dog to Standing: A New Way to Morph 3D Shapes
MCML Research Insight - With Lu Sang and Daniel Cremers
Ever wondered how a 3D shape can smoothly change — like a robot arm bending or a dog rising from sitting to standing — without complex simulations or hand-crafted data? Researchers from MCML and the University of Bonn tackled this challenge in their ICLR 2025 paper, “Implicit Neural Surface Deformation with Explicit Velocity Fields”.
04.12.2025
When to Say “I’m Not Sure”: Making Language Models More Self-Aware
MCML Research Insight - With Yawei Li, David Rügamer, Bernd Bischl, and Mina Rezaei
Large language models like ChatGPT or Gemini are now everywhere, from summarizing text to writing code or answering simple questions. But there’s one thing they still struggle with: admitting uncertainty. Ask a fine-tuned LLM a tricky question, and it might sound quite confident, even when it’s completely wrong. This “overconfidence” …
©MCML
01.12.2025
Research Stay at Princeton University
Abdurahman Maarouf – Funded by the MCML AI X-Change Program
From May to July, I spent three exciting months as a visiting researcher at the Computer Science Department of Princeton University, hosted by Prof. Manoel Horta Ribeiro. The visit grew out of a keynote Manoel gave at LMU. After his talk, we discussed potential joint projects at the intersection of causal inference, machine learning, and social …