Deep Graph Spectral Evolution Networks for Graph Topological Evolution

Authors: Negar Etemadyrad, Qingzhe Li, Liang Zhao7358-7366

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

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on multiple synthetic and real-world datasets demonstrate outstanding performance.
Researcher Affiliation Academia 1 George Mason University, Fairfax, VA, 22033 2 Emory University, Atlanta, GA, 30307
Pseudocode No The paper includes a diagram (Figure 1) illustrating the network architecture, but no pseudocode or clearly labeled algorithm blocks.
Open Source Code Yes The proposed method is implemented with Pytorch deep learning framework.1 1https://github.com/netemady/GSEN
Open Datasets Yes Real-world HCP Dataset: ...obtained from the human connectome project (HCP) (Van Essen et al. 2013) 2. 2http://www.humanconnectomeproject.org/
Dataset Splits Yes For all comparison and our methods, 5-fold cross-validation is performed, where for each run we select one subset as test and the remaining 4 as training set. In the training set, 20% is randomly selected as validation to determine hyperparameters through a grid search.
Hardware Specification Yes conducted on a 64-bit machine with 40 GB memory, a 4-core Intel CPU and an Nvidia RTX-2080 Ti GPU.
Software Dependencies No The proposed method is implemented with Pytorch deep learning framework. No specific version number for PyTorch is provided.
Experiment Setup No The parameter settings of GT-GAN, C-DGT, and our method are detailed in supplementary materials, where we also included key parameter sensitivity analyses. These details are not in the main text.