Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The Shift Toward Open and Reproducible AI Research," Under Review, 2026. doi: 10.48550/arXiv.2606.16974..
Rethinking Explaining Graph Neural Networks via Non-parametric Subgraph Matching
Authors: Fang Wu, Siyuan Li, Xurui Jin, Yinghui Jiang, Dragomir Radev, Zhangming Niu, Stan Z. Li
ICML 2023 | Venue PDF | LLM Run Details | Input Tokens: 21,832 Total number of tokens sent to the LLM as input for this paper's analysis. | Output Tokens: 5,506 Total number of tokens produced by the LLM (including reasoning/thinking tokens) for this paper's analysis.
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on synthetic and real-world datasets show the effectiveness of our Match Explainer by outperforming all state-of-the-art parametric baselines with significant margins. Results also demonstrate that Match Drop is a general scheme to be equipped with GNNs for enhanced performance. |
| Researcher Affiliation | Collaboration | 1School of Engineering, Westlake University, Hangzhou, China 2Mindrank AI, Hangzhou, China 3Department of Computer Science, Yale University, New Haven, United States. |
| Pseudocode | Yes | Algorithm 1 Workflow of Match Explainer |
| Open Source Code | Yes | The code is available at https://github. com/smiles724/Match Explainer. |
| Open Datasets | Yes | Following Wang et al. (2021b), we use four standard datasets... MUTAG (Debnath et al., 1991; Kazius et al., 2005)... BA-3Motif... MNIST... VG-5 (Pope et al., 2019; Krishna et al., 2017). |
| Dataset Splits | No | The paper mentions 'full training and validation data as the reference set' and 'testing accuracy' but does not provide specific percentages or counts for training, validation, and test splits. |
| Hardware Specification | Yes | All experiments are conducted on a single A100 PCIE GPU (40GB). |
| Software Dependencies | No | The paper mentions using 'Adam optimizer' but does not provide specific version numbers for software dependencies such as programming languages, libraries, or frameworks. |
| Experiment Setup | Yes | Regarding the re-implementation of Refine in BA-3Motif, we use the original code with the same hyperparameters, and we adopt Adam optimizer (Kingma & Ba, 2014) and set the learning rate of pre-training and fine-tuning as 1e-3 and 1e-4, respectively. |