Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
Q: Fachverband Quantenoptik und Photonik
Q 74: Quantum Information – Concepts and Methods
Q 74.4: Vortrag
Freitag, 6. März 2026, 11:45–12:00, P 10
Exploring Disorder Effects in Quantum Generative Models — •Nikolaos Palaiodimopoulos1,2, Yannick Werner1,2, Jasmin Frkatovic1, Vitor Fortes Rey1, Matthias Tchöpe2, Sungho Suh2, Paul Lukowicz1,2, and Maximilian Kiefer- Emmanouilidis1,2 — 1RPTU Kaiserslautern-Landau — 2DFKI Kaiserslautern
Disordered quantum many-body systems (DQS) and quantum neural networks (QNNs) exhibit strong structural parallels, with a DQS effectively functioning as a QNN with randomly initialized parameters. We show that random processes can act as a deceptive quantum generative mechanism in QNNs, where unitarity preserves memory effects absent in classical networks. These effects impact both the learnability and trainability of QNNs and can lead to an overestimation of their generative capabilities. While DQS can be useful for tasks such as image augmentation, we caution that evaluations on overly simple datasets may misrepresent the true power of current quantum generative models.
Keywords: Disordered Quantum Systems; Quantum Machine Learning; Generative Models; Quantum Neural Networks; Many-body physics