![[2512.17762] Can You Hear Me Now? A Benchmark for Long-Range Graph Propagation](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2512.17762] Can You Hear Me Now? A Benchmark for Long-Range Graph Propagation
Summary
This article introduces ECHO, a benchmark for evaluating long-range graph propagation in graph neural networks (GNNs), addressing a critical challenge in machine learning.
Why It Matters
Long-range interactions in GNNs are crucial for various scientific applications. The ECHO benchmark provides a systematic approach to assess GNN performance, highlighting gaps and potential improvements, which is essential for advancing AI in scientific research.
Key Takeaways
- ECHO benchmark evaluates GNNs on long-range graph propagation tasks.
- Includes synthetic tasks and real-world datasets for comprehensive assessment.
- Identifies performance gaps in existing GNN architectures, emphasizing design choices for improvement.
- Highlights the importance of long-range interactions in scientific applications.
- Sets a new standard for evaluating GNN capabilities in AI for science.
Computer Science > Machine Learning arXiv:2512.17762 (cs) [Submitted on 19 Dec 2025 (v1), last revised 23 Feb 2026 (this version, v2)] Title:Can You Hear Me Now? A Benchmark for Long-Range Graph Propagation Authors:Luca Miglior, Matteo Tolloso, Alessio Gravina, Davide Bacciu View a PDF of the paper titled Can You Hear Me Now? A Benchmark for Long-Range Graph Propagation, by Luca Miglior and Matteo Tolloso and Alessio Gravina and Davide Bacciu View PDF HTML (experimental) Abstract:Effectively capturing long-range interactions remains a fundamental yet unresolved challenge in graph neural network (GNN) research, critical for applications across diverse fields of science. To systematically address this, we introduce ECHO (Evaluating Communication over long HOps), a novel benchmark specifically designed to rigorously assess the capabilities of GNNs in handling very long-range graph propagation. ECHO includes three synthetic graph tasks, namely single-source shortest paths, node eccentricity, and graph diameter, each constructed over diverse and structurally challenging topologies intentionally designed to introduce significant information bottlenecks. ECHO also includes two real-world datasets, ECHO-Charge and ECHO-Energy, which define chemically grounded benchmarks for predicting atomic partial charges and molecular total energies, respectively, with reference computations obtained at the density functional theory (DFT) level. Both tasks inherently depend on capturing complex...