On Self-Distilling Graph Neural Network

Authors: Yuzhao Chen, Yatao Bian, Xi Xiao, Yu Rong, Tingyang Xu, Junzhou Huang

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

Reproducibility Variable Result LLM Response
Research Type Experimental 4 Experiments
Researcher Affiliation Collaboration Yuzhao Chen1,2, , Yatao Bian2, , Xi Xiao1,3, , Yu Rong2 , Tingyang Xu2 , Junzhou Huang2,4 1Tsinghua University 2Tencent AI Lab 3Peng Cheng Laboratory 4University of Texas at Arlington
Pseudocode No The paper does not contain any pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any statement or link regarding the release of source code.
Open Datasets Yes Node Classification Table 2 summarizes the results of GNNs with various depths on Cora, Citeseer and Pub Med [2008]. Graph Classification Table 3 summarizes the results of various popular GNNs on the graph kernel classification datasets, including ENZYMES, DD, and PROTEINS in TU dataset [2016].
Dataset Splits Yes We follow the setting of semi-supervised learning, using the fixed 20 nodes per class for training. [...] each experiment is conducted by 10-fold cross validation with the splits ratio at 8:1:1 for training, validating and testing.
Hardware Specification No The paper does not specify any hardware details (e.g., GPU/CPU models, memory) used for running experiments.
Software Dependencies No The paper does not specify any software dependencies with version numbers.
Experiment Setup Yes For our method, the hyper-parameters of α and β are both set to 0.01 and γ is 0. [...] The hyper-parameters of γ is fixed to 0 for node classification, and we determine α and β via a simple grid search. Details are provided in Appendix. [...] Hyper-parameter settings are deferred to Appendix.