How to run the code:
baseline.ipynb: Code for baseline linear classifier. Simply run the whole notebook and it will do a grid search on hyper-parameters and print out the results. Need to change the path to data.csv before running the notebook.
problog_model.ipynb: Code for our problog model. Simply run the whole notebook and it will do the following: (Need to change the path to data.csv before running the notebook.)
- generate the .pl file for our problog model
- run query on all songs and evaluate using the metrics: root mean squared error (rmse) and mean absolute error (mae). Note: learning and inference are computationally intractable for large dataset. To run a toy example, we recommand choosing a dataset size < 30
- investiage the growth of learning and inference time as dataset size increases and plot the results
main.py and hgnn:
Code for RGCN. To run hyperparameter search python main.py -m RGCN -t node_classification -d fma -g 0 --use_hpo. To run with the best hyperparameters we found python main.py -m RGCN -t node_classification -d fma -g 0 --use_best_config
BN_model.py: Require dependency on PyTAC libray
gen_BN_from_dataset(CSV_PATH, DATA_SIZE): generate complete bayesian network from dataset, but learning is intractable train_toy_BN_from_dataset(): make simple BN baseline and learn from dataset
.