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 embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Local-Global Defense against Unsupervised Adversarial Attacks on Graphs
Authors: Di Jin, Bingdao Feng, Siqi Guo, Xiaobao Wang, Jianguo Wei, Zhen Wang
AAAI 2023 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments demonstrate that our strategies can enhance the robustness of representation against various adversarial attacks on three benchmark graphs. |
| Researcher Affiliation | Academia | Di Jin1, Bingdao Feng1, Siqi Guo1, Xiaobao Wang1*, Jianguo Wei1, Zhen Wang2 1College of Intelligence and Computing, Tianjin University, Tianjin, China 2School of Cybersecurity, Northwestern Polytechnical University, Xi an, Shaanxi, China |
| Pseudocode | Yes | Algorithm 1: Optimization algorithm |
| Open Source Code | No | The paper does not provide any explicit statement or link regarding the public availability of its source code. |
| Open Datasets | Yes | We use three real-world datasets in our experiments, i.e., Cora, Citeseer (Sen et al. 2008) and Polblogs (Adamic and Glance 2005). |
| Dataset Splits | Yes | For Cora and Citeseer, randomly assign them to the training, verification, and test sets in the ratio of 1:1:8. |
| Hardware Specification | No | The paper does not explicitly describe the hardware used to run its experiments. |
| Software Dependencies | No | The paper mentions various models and frameworks (e.g., GCN, GAT, DGI) but does not specify version numbers for any software dependencies. |
| Experiment Setup | Yes | For our model, we set the parameters h = 0.2, α = 1 and β = 0.4. At the stage of evaluating, we consider both the performance and the robustness of the model, so we employ the four attack methods indicated above, and set different perturbation ratios with a step of 10%. All comparative learning baselines use a two-layer GCN as the encoder and use the default setting. |