Search papers, labs, and topics across Lattice.
This paper benchmarks classical and quantum-oriented node embeddings for graph classification using a unified pipeline across five TU datasets and a converted QM9 dataset. They find that quantum-oriented embeddings, including circuit-defined variational embeddings and quantum-inspired embeddings, offer consistent gains on structure-driven benchmarks. The study elucidates practical trade-offs between inductive bias, trainability, and stability under a fixed training budget when selecting node embeddings.
Quantum-inspired node embeddings can outperform classical baselines in graph neural networks, but only on structure-driven datasets, revealing critical dataset dependencies.
Node embeddings act as the information interface for graph neural networks, yet their empirical impact is often reported under mismatched backbones, splits, and training budgets. This paper provides a controlled benchmark of embedding choices for graph classification, comparing classical baselines with quantum-oriented node representations under a unified pipeline. We evaluate two classical baselines alongside quantum-oriented alternatives, including a circuit-defined variational embedding and quantum-inspired embeddings computed via graph operators and linear-algebraic constructions. All variants are trained and tested with the same backbone, stratified splits, identical optimization and early stopping, and consistent metrics. Experiments on five different TU datasets and on QM9 converted to classification via target binning show clear dataset dependence: quantum-oriented embeddings yield the most consistent gains on structure-driven benchmarks, while social graphs with limited node attributes remain well served by classical baselines. The study highlights practical trade-offs between inductive bias, trainability, and stability under a fixed training budget, and offers a reproducible reference point for selecting quantum-oriented embeddings in graph learning.
