signed_graph_convolutional_network_reproduction

Reproduction of paper Signed Graph Convolutional Network

The result for tables 6 and 7 in our report can be found in file runs/res_2022-01-24_13-45-46.csv

Contents

data: includes the datasets we used

runs: includes the logs of our experiment runs (Experiment 1)

runs_dist: includes the logs of our statistical significance analysis (Experiment 2)

runs_lambda: includes the logs of our hyperparameter sensitivity analysis (Experiment 3)

src: contains our models, the training and evaluation pipeline

Running the Code

Environment

We provide a conda environment.yml file which includes the environment which we used.

Set it up with this command:

conda env create -f environment.yml

Experiments

To run Experiment 1, simply run main.py. For the other experiments, you can adjust the specific independent variables (sgcn2_hyperpars_torch["lamb"], REPEATS) in main.py.