Research map / Graph learning / GNN architectures
Neutral-atom quantum features as complementary structural encodings for graph learning
Igor Sokolov, Mathieu Garrigues, Matthias Hecker, Lorenzo Moro, Shaheen Acheche
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes using measurement features from neutral-atom quantum processors running graph-dependent Rydberg Hamiltonians as complementary structural encodings for GNNs.
Problem
Classical structural encodings for GNNs (random walks, Laplacian spectra) capture only specific structural statistics of a graph, leaving room for alternative graph descriptors.
Approach
Neutral-atom QPUs realize graph-dependent Rydberg Hamiltonians, so quantum measurement-derived features probing graph topology are used as structural encodings complementary to classical ones for graph learning.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add AFM quantum features to RRWP | RRWP encoding, DeepSet probe · ZINC intrinsic probe: degree | MAE (x1e-2): 14.4 -> 1.18 | Improved |
| add AFM quantum features to RRWP | RRWP encoding, DeepSet probe · ZINC intrinsic probe: node min cycle | MAE (x1e-2): 0.73 -> 0.36 | Improved |
| add FM to RRWP+AFM | RRWP+AFM · ZINC intrinsic probe: eccentricity | MAE: 1.77 -> 1.47 (eccentricity); best on local tasks otherwise | Mixed |
| add quantum features to RRWP in GNN backbones (champion hybrid vs RRWP) | six GNN backbones · MUTAG and ZINC subset | F1 / MAE: ties or improves in 10 of 12 cells; sign test p = 0.019 | Improved |
| add AFM_stat+FM_stat to RRWP | GAT · ZINC subset | MAE: 0.223 -> 0.209 | Improved |
| add quantum features to RRWP on GeneralConv | GeneralConv · ZINC subset | MAE: 0.145 vs 0.145 (RRWP+Laplacian) | No clear effect |
| use quantum features alone instead of RRWP (ablation) | GNN backbones · MUTAG and ZINC subset | F1 / MAE: ZINC GT MAE 0.113 -> 0.163 (AFM) | Hurt |
| replace emulated with QPU hardware features (AFM_stat hybrid) | RRWP + AFM_stat · MUTAG | macro F1: within 0.01 F1 on all six backbones | No clear effect |
| replace emulated with hardware AFM features (hybrid) | RRWP + AFM · MUTAG | macro F1: GCN 0.878 -> 0.837; within 0.014 on five of six | Mixed |
| use gated fusion instead of concatenation (ablation) | ESA with RWSE*+quantum features · MUTAG | binary F1: RWSE*+AFM+FM 0.903 -> 0.922 | Improved |
| concatenate quantum features to RWSE* in ESA | ESA · MUTAG | binary F1: 0.905 -> 0.899 (AFM_stat) | No clear effect |
| use AFM+FM quantum features instead of classical features on ladder graphs | logistic regression probe · Synthetic ladder graphs | accuracy: 0.641 -> 0.875 | Improved |
Experimental setup
Baselines: random walk encodings Laplacian spectral encodings
Builds on: random walk positional encoding Laplacian positional encoding
Closest papers by meaning
- Quantum Positional Encodings for Graph Neural Networks (ICML 2024 · similarity 0.88)
- HSG-12M: A Large-Scale Benchmark of Spatial Multigraphs from the Energy Spectra of Non-Hermitian Crystals (ICLR 2026 · similarity 0.87)
- Learning the Electronic Hamiltonian of Large Atomic Structures (ICML 2025 · similarity 0.86)
- Neural QAOA$^2$: Differentiable Joint Graph Partitioning and Parameter Initialization for Quantum Combinatorial Optimization (ICML 2026 · similarity 0.86)
- QBMK: Quantum-based Matching Kernels for Un-attributed Graphs (ICML 2024 · similarity 0.84)
- Topological data analysis on noisy quantum computers (ICLR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
