WaveNet: Tackling Non-stationary Graph Signals via Graph Spectral Wavelets
Authors: Zhirui Yang, Yulan Hu, Sheng Ouyang, Jingyu Liu, Shuqiang Wang, Xibo Ma, Wenhan Wang, Hanjing Su, Yong Liu
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We also conduct node classification experiments on real-world graph benchmarks and achieve superior performance on most datasets. |
| Researcher Affiliation | Collaboration | Zhirui Yang1 , Yulan Hu1,2 , Sheng Ouyang1,2 , Jingyu Liu1 , Shuqiang Wang3 , Xibo Ma4,* , Wenhan Wang5 , Hanjing Su5 , Yong Liu1,* 1Renmin University of China 2School of Artificial Intelligence, University of Chinese Academy of Sciences 3Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences 4Institute of Automation, Chinese Academy of Sciences (CASIA) 5Tencent Inc. |
| Pseudocode | No | The paper describes the architecture details in Appendix A.3 but does not contain a structured pseudocode or algorithm block. |
| Open Source Code | Yes | Our code is available at https:// github.com/Bufordyang/Wave Net |
| Open Datasets | Yes | Data: We conduct node classification task on real-world datasets following (He et al. 2021). Including three citation network Cora, Cite Seer and Pub Med (Sen et al. 2008; Yang, Cohen, and Salakhudinov 2016) and the Amazon co-purchase graph Computers and Photo (Mc Auley et al. 2015).We also include the Wikipedia graph Chameleon and Squirrel (Rozemberczki, Allen, and Sarkar 2021), the Actor co-occurrence graph is attained from (Pei et al. 2020). |
| Dataset Splits | Yes | Setup: We adopt the randomly splitting the node set into train/validation/test set with ratio 60%/20%/20%, and conduct full-supervised node classification task with each baseline model. |
| Hardware Specification | No | The paper does not explicitly describe the specific hardware used for running its experiments (e.g., GPU models, CPU types, or memory specifications). |
| Software Dependencies | No | The paper mentions 'Pytorch Geometric library' and 'tensorflow package' but does not specify their version numbers for reproducibility. |
| Experiment Setup | Yes | Setup: We adopt the randomly splitting the node set into train/validation/test set with ratio 60%/20%/20%, and conduct full-supervised node classification task with each baseline model. For fairness, we also use the same evaluate setting as fowling (He et al. 2021): we generate 10 random splits by random seeds and evaluate all models on the same splits, and report the average metric for each model. ... For fairness, the order of polynomial filter K = 10 and the number of wavelet bases as well. |