Quantum-inspired interferometry for graph structure learning
by
Albano 3: 5230 - Xenon (12 seats)
Albano Building 3
This talk introduces a quantum-inspired method for structure learning on graphs using simulated interferometry. By modeling Schrödinger-type wave propagation over network topologies, we obtain time-resolved complex wavefunctions whose interference patterns encode local and global structure. These frequency-resolved responses are assembled into training data for a neural embedding model, which recovers low-dimensional spatial representations constrained by graph connectivity and signal similarity. Current work seeks to use this method to obtain superior error correction and boundary-detecting properties compared to classical random-walk based methods. We discuss ongoing work and potential applications to reconstructing DNA barcode networks for spatial transcriptomics.