Mainz 2026 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
Q: Fachverband Quantenoptik und Photonik
Q 67: Poster – Quantum Information
Q 67.25: Poster
Thursday, March 5, 2026, 17:00–19:00, Philo 2. OG
Quantum-Inspired Low-Entanglement Optimization Techniques for Image Segmentation — •Marie Gogolin1, Richard Castro1,2, Yannick Werner1,3, Ali Mogisheh2, Alexander Geng2, Arcesio Castaneda Medina2, Paul Lukowicz1,3, and Maximilian Kiefer-Emmanouilidis1,3 — 1RPTU Kaiserslautern-Landau — 2Fraunhofer ITWM Kaiserslautern — 3DFKI Kaiserslautern
Optimization problems such as image segmentation can be mapped to ground-state computations in Ising-type models, allowing quantum and quantum-inspired methods to be applied to real-world tasks. In this work, we investigate a spectrum of approaches to solving such problems, starting from the Max-Cut formulation and comparing exact diagonalization, imaginary-time evolution, and quantum annealing techniques. The focus lies on tensor-network methods. Matrix Product States and Projected Entangled Pair States, which enable scalable simulations of one- and two-dimensional systems with controllable entanglement. We further integrate quantum annealing concepts with tensor-network optimization by combining generalized PEPS using the simple-update scheme with a Floquet-based adiabatic evolution approximation. Numerical experiments on image segmentation benchmark the strengths and limitations of these techniques across varying system sizes and problem structures. The results demonstrate the potential of low-entanglement tensor networks as powerful tools for quantum-inspired optimization, bridging theoretical developments with practical applications in computer vision.
Keywords: Quantum Inspired Methods; Tensornetworks; Quantum Image Segmentation