UAG: Uncertainty-aware Attention Graph Neural Network for Defending Adversarial Attacks

Authors: Boyuan Feng, Yuke Wang, Yufei Ding7404-7412

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Intensive experiments show that our proposed defense approach outperforms the state-of-the-art solutions by a significant margin. Rigorous experiments and studies across various datasets on mainstream GNNs show that our proposed defense approach outperforms the state-of-the-art RGCN by a significant margin.
Researcher Affiliation Academia Boyuan Feng, Yuke Wang, Yufei Ding University of California, Santa Barbara {boyuan, yuke_wang, yufeiding}@ucsb.edu
Pseudocode No The paper describes the methodology using prose and mathematical equations but does not include explicit pseudocode or algorithm blocks.
Open Source Code No The paper does not explicitly state that source code for the described methodology is publicly available, nor does it provide a link to a code repository.
Open Datasets Yes We select the most typical datasets (Cora, Citeseer, and Pubmed) used by many GNN papers (Kipf and Welling 2017; Xu et al. 2019b; Hamilton, Ying, and Leskovec 2017) to evaluate our UAG.
Dataset Splits Yes We follow the common data split by selecting 10% nodes as the training dataset, 10% nodes as the validation dataset, and 80% nodes as the testing dataset.
Hardware Specification Yes We evaluate UAG on a Dell Workstation T7910 (Ubuntu 18.04) with an Intel Xeon CPU E5-2603, 64 GB memory, and an NVIDIA 1080Ti GPU with 12GB memory.
Software Dependencies No We implement UAG based on Py Torch Geometric (Fey and Lenssen 2019). No specific version numbers for PyTorch Geometric or other software dependencies are provided.
Experiment Setup Yes In our experiments, we can achieve satisfying results by using ατ = 15 and βτ = 2.5.