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 [1].
Graph Matching with Partially-Correct Seeds
Authors: Liren Yu, Jiaming Xu, Xiaojun Lin
JMLR 2021 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Numerical experiments corroborate our theoretical findings, demonstrating the superiority of our 2hop algorithm on a variety of synthetic and real graphs. |
| Researcher Affiliation | Academia | Liren Yu EMAIL Elmore Family School of Electrical and Computer Engineering Purdue University West Lafayette, IN 47907, USA Jiaming Xu EMAIL The Fuqua School of Business Duke University Durham, NC 27708, USA Xiaojun Lin EMAIL Elmore Family School of Electrical and Computer Engineering Purdue University West Lafayette, IN 47907, USA |
| Pseudocode | Yes | Algorithm 1 Graph Matching based on Counting j-hop Witnesses. 1: Input: G1, G2, π, j 2: Generate j-hop adjacency matrices Aj and Bj based on G1 and G2, and Π based on π; 3: Output eπ = GMWM(Wj), where Wj = AjΠBj. |
| Open Source Code | Yes | Our code has been released on Git Hub at https://github.com/Leron33/Graph-matching. |
| Open Datasets | Yes | We use a Facebook friendship network of 11621 students and staffs from Standford university provided in Traud et al. (2012) as the parent graph G0. We use the Autonomous Systems (AS) data set from Leskovec and Krevl (2014) to test the graph matching performance on real graphs. Their experiment is carried on the SHREC 16 data set in L ahner et al. (2016). We follow the Princeton benchmark protocol in Kim et al. (2011) to evaluate the matching quality. |
| Dataset Splits | Yes | The SHREC 16 data set provides 25 deformable 3D shapes (15 for training and 10 for testing) undergoing different topological changes. |
| Hardware Specification | Yes | The computational environment is MATLAB R2017a on a standard PC with 2.4 GHz CPU and 8 GB RAM. |
| Software Dependencies | Yes | The computational environment is MATLAB R2017a on a standard PC with 2.4 GHz CPU and 8 GB RAM. |
| Experiment Setup | Yes | In Fig. 6, we fix n = 10000 and s = 0.9, and plot the accuracy rates for p = n 3/4 and p = n 6/7. ... We run the three algorithms with a given number of iterations L = 0, 1, 2. In Fig. 7, we consider the same setup as in Fig. 6. ... For the Noisy Seeds algorithm, we choose the threshold r = 3 for p = n 3/4 and r = 2 for p = n 6/7. |