At CIRCLE, we aim to make machines see and perceive the world—not just to recognize pixels, but to understand the structure, symmetry, and meaning embedded in the physical and abstract world. We explore the frontiers of perception, learning, and geometry, blending deep learning with the mathematical elegance of topology, differential geometry, and even quantum theory.
Our philosophy is holistic: we believe that intelligent systems must be as deeply rooted in theoretical understanding as they are in real-world utility. We let applications inform learning, and learning theories guide applications. This dual lens is what makes our work both principled and impactful. We are driven by curiosity and a commitment to open, inclusive, transformative science.
Our research spans:
- 3D Computer Vision: Perceiving and generating structure, motion, and semantics in three and four-dimensional spaces
- Geometric & Topological Deep Learning: Learning on complex domains: graphs, manifolds, cell complexes and beyond
- Learning Theory:Statistical & topological foundations of machine learning and mechanistic interpretability
- Quantum Computer Vision and Machine Learning: Novel computational paradigms towards vision and learning
- AI-driven for Biochemistry: Designing molecules and proteins with symmetry and structure-aware higher-order priors
Join us as we reimagine machine intelligence from the ground up—with rigor, creativity, and curiosity about the world and how we learn to see it.
Highlights
Our Publications
We actively publish at the top venues of computer vision and machine learning, including CVPR, NeurIPS, ICLR, ICCV, ECCV, ICML and 3DV.
Our Research
We publish in a variety of areas that cross-feed each other, including: 3D Computer Vision including 3D/4D Generative Priors, Geometric & Topological Deep Learning, Statistical & Topological Learning Theory, Mechanistic Interpretability, Quantum Computer Vision and Generative Models for Biochemistry.
Our funders
