Quantum Doubly Stochastic Transformers
- Jannis Born
- Filip Skogh
- et al.
- 2025
- NeurIPS 2025
Bringing Artificial Intelligence and Quantum Computing together opens up new ways to design algorithms that make practical use of today's quantum hardware, while also anticipating future fault-tolerant machines.
By combining algorithmic innovation from classical and quantum information theory with modern AI methods, we aim to remove computational bottlenecks in existing algorithms for high-dimensional, compute-intensive problems. Building on our work in foundational models for science, we focus on hybrid approaches that leverage currently available quantum devices, helping to connect theoretical advances with practical implementations. This includes addressing concrete challenges such as eigenvalue problems, subspace identification, and modeling in both deterministic and probabilistic settings, with applications ranging from industry and materials science to complex system simulations.
Ultimately, our goal is to accelerate discovery and innovation by creating practical computational tools that blend the strengths of AI and Quantum Computing, and that can evolve alongside rapidly advancing hardware and software ecosystems.