Adaptive Graph Convolutional Neural Networks
Authors: Ruoyu Li, Sheng Wang, Feiyun Zhu, Junzhou Huang
AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on nine graph-structured datasets have demonstrated the superior performance improvement on both convergence speed and predictive accuracy. |
| Researcher Affiliation | Collaboration | The University of Texas at Arlington, Arlington, TX 76019, USA Tencent AI Lab, Shenzhen, 518057, China |
| Pseudocode | Yes | Algorithm 1 SGC-LL Layer |
| Open Source Code | No | The paper does not provide any explicit statement about open-source code availability or a link to a code repository. |
| Open Datasets | Yes | Delaney Dataset (Delaney 2004) contains aequeous solubility data for 1,144 low molecular weight compounds. The largest compound in the dataset has 492 atoms, while the smallest only consists of 3 atoms. NCI Database has around 20,000 compounds and 60 prediction tasks from drug reaction experiments to clinical pharmacology studies. At last, Az-log D dataset from ADME (Vugmeyster, Harrold, and Xu 2012) offers the log D measurements on permeability for 4200 compounds. Besides, we also have a small dataset of 642 compounds for hydration-free energy study. The presented task-averaged RMSE scores and standard deviations were obtained after 5-fold cross-validation. Tox21 Dataset (Mayr et al. 2016) contains 7,950 chemical compounds and labels for classifications on 12 essays of toxicity. Clin Tox is a public dataset of 1451 chemical compounds for clinical toxicological study together with labels for 2 tasks. Sider (Kuhn et al. 2010) database records 1392 drugs and their 27 different side effects or adverse reactions. Toxcast (Dix et al. 2006) is another toxicological research database that has 8,599 SMILES together with labels for 617 predictive tasks. Sydney urban point cloud dataset contains street objects scanned with a Velodyne HDL-64E LIDAR, collected in the CBD of Sydney, Australia. |
| Dataset Splits | Yes | The presented task-averaged RMSE scores and standard deviations were obtained after 5-fold cross-validation. Table 2: Task-averaged ROC-AUC Scores on Tox21, Clin Tox, Sider, Toxcast Datasets. The same benchmarks as Table. 1. |
| Hardware Specification | No | The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory) used for running the experiments. |
| Software Dependencies | No | The paper does not specify any software dependencies with version numbers (e.g., programming languages, libraries, frameworks) used for the experiments. |
| Experiment Setup | No | The paper describes the network configuration and general training aspects (e.g., batch normalization, max pooling, residual graph learning), but it does not provide specific hyperparameter values (e.g., learning rate, batch size, number of epochs) or detailed optimizer settings needed for full reproducibility. |