Adaptive Kernel Graph Neural Network

Authors: Mingxuan Ju, Shifu Hou, Yujie Fan, Jianan Zhao, Yanfang Ye, Liang Zhao7051-7058

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments are conducted on acknowledged benchmark datasets and promising results demonstrate the outstanding performance of our proposed AKGNN by comparison with state-of-the-art GNNs.
Researcher Affiliation Academia 1 University of Notre Dame, Notre Dame, IN 46556 2 Case Western Reserve University, Cleveland, OH 44106 3 Emory University, Atlanta, GA 30322
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code Yes The source code is publicly available at: https://github.com/jumxglhf/AKGNN.
Open Datasets Yes The three datasets we evaluate are Cora, Citeseer and Pubmed (Sen et al. 2008).
Dataset Splits Yes (e.g., publicly fixed 20 nodes per class for training, 500 nodes for validation, and 1,000 nodes for testing).
Hardware Specification Yes All the experiments in this work are implemented on a single NVIDIA Ge Force RTX 2080 Ti with 11 GB memory size and we didn t encounter any memory bottleneck issue while running all experiments.
Software Dependencies No The paper mentions 'We utilize Py Torch as our deep learning framework to implement AKGNN' but does not specify the version number for PyTorch or any other software dependencies.
Experiment Setup Yes The number of layers K is 5, the hidden size d(K) is 64, the dropout rate between propagation layers is 0.6, the learning rate is 0.01, the weight decay rate is 5e-4, and patience for early stopping is 100 iterations.