configulation_file.md

Configulation file

"model.py"

model python script

"dataset"

dataset joblib file

"validation_dataset"

validation dataset joblib file

"validation_data_rate"

generating validation dataset by splitting training dataset with validation_data_rate

"epoch"

the maximum numeber of epochs

"batch_size"

the number of samples in a minibatch

"patience"

patience parameter for early stopping

"learning_rate"

(initial) learning rate

"shuffle_data"

shuffling data after loading dataset

"with_feature"

In GCN, a node has feature or not.

*"with_node_embedding"

In GCN, a node has embedding vector or not.

"embedding_dim"

When with_node_embedding=True, The dimension of an embedding vector.

"normalize_adj_flag"

enables normalization of adjacency matrices

"split_adj_flag"

enables splitting adjacency matrices using dgree of a node

"order"

order of adjacency matrices

"param"

optional parameters for neural network archtecture (used in Baysian optimization)

"k-fold_num"

specifies the number of folds related to train_cv command.

"save_interval"

inter

"save_model_path"

path to save model

"save_result_train"

csv file name to save summarized results (train command)

"save_result_valid"

csv file name to save summarized results (train command)

"save_result_test"

csv file name to save summarized results (infer command)

"save_result_cv"

json file name to save summarized results (train_cv command)

"save_info_train"

json file name to save detailed information (train command)

"save_info_valid"

json file name to save detailed information (train command)

"save_info_test"

json file name to save detailed information (infer command)

"save_info_cv"

json file name to save cross-validation information (train_cv command)

"make_plot"

enables plotting results

"plot_path"

path to save plot data

"plot_multitask"

plotting results of multitaslk

"profile"

for profiling using the tensorflow profiler

stratified_kfold

for using stratified k-fold