TSFL

Requirements

• Python 3.6
• pytorch 1.9.0
• torchvision 0.9.0
• tensorflow 1.15.0
• Keras 2.3.1
• numpy 1.15.4
• scipy 1.1.0
• scikit-learn 0.19.1
• sklearn 0.19.1
• annoy 1.17.0
• h5py 2.10.0

How to run

environment

conda create -n gnn python=3.6
bash install_env.sh

How to run

• 1. Get Dataset:

  You can download ISRUC-Sleep-S3 dataset by the following url http://dataset.isr.uc.pt/ISRUC_Sleep/subgroupIII put all data in "./data/ISRUC_S3/RawData"

  http://dataset.isr.uc.pt/ISRUC_Sleep/ExtractedChannels/subgroupIII-Extractedchannels put all data in "./data/ISRUC_S3/ExtractedChannels"

• 2. Data preparation:

  To facilitate reading, we preprocess the dataset into a single .npz file:

  python preprocess.py

  In addition, distance based adjacency matrix is provided at ./data/ISRUC_S3/DistanceMatrix.npy.

• 3. Configurations of feature extraction module:

  Write the config file in the format of the example.

  We provide a config file at ./config/ISRUC.config

• 4. Prepare feature extraction module:

  Run python train_FeatureNet.py with -c and -g parameters. After this step, the features learned by a feature net will be stored.

  • -c: The configuration file.
  • -g: The number of the GPU to use. E.g.,0,1,3. Set this to-1 if only CPU is used.

  python train_FeatureNet.py -c ./config/ISRUC.config -g 0

• 5. Experiment:

  Get back to thr Root folder. Run python gnn_experiment.py.
  In this command, 3 arguments can be changed.

  • --model: which GNN model you use. Currently supports sage, gat and gcn.

  • --case_name: which connective function you use. Currently supports distance, knn, pcc and plv.

  • --data_dir: which dir to save experiment logs, model and temporary files.

    single machine experiment:

  python gnn_experiment.py --model gat --case_name knn --data_dir ./result/ISRUC_S3_knn/
  python gnn_experiment.py --model gat --case_name plv --data_dir ./result/ISRUC_S3_plv/
  python gnn_experiment.py --model gat --case_name distance --data_dir ./result/ISRUC_S3_distance/
  python gnn_experiment.py --model gat --case_name pcc --data_dir ./result/ISRUC_S3_distance/

  multiple machines experiment:

  python fed_experiment.py --model gat --case_name knn --data_dir ./result/ISRUC_S3_knn  
  python fed_experiment.py --model gat --case_name plv --data_dir ./result/ISRUC_S3_plv  
  python fed_experiment.py --model gat --case_name distance --data_dir ./result/ISRUC_S3_distance  
  python fed_experiment.py --model gat --case_name pcc --data_dir ./result/ISRUC_S3_pcc  
  

Summary of commands to run:

./get_ISRUC_S3.sh
python preprocess.py
python train_FeatureNet.py -c ./config/ISRUC.config -g 0
python fed_experiment.py --model gat --case_name knn --data_dir ./result/ISRUC_S3_knn