DTI-MACF

Environment Settings

• Python == 3.6.13
• pytorch == 1.7.1
• numpy == 1.19.5
• scikit-learn == 0.24.2

Parameter Settings

• epochs: the number of epochs to train
• lr: learning rate
• embed_dim: embedding dimension
• N: a parameter of L0, the default is the number of triples
• droprate: dropout rate
• batch_size: batch size for training

Data Process

Datasets are available in datasets/

The raw data is processed into bipartite graph through preprocess.py as input data

python preprocess.py

Basic Usage

Both training and testing procedures can be achived by the scriptrun.py

python run.py 

Hyper-parameters Tuning

There are three key hyper-parameters: number of components, lr and embed_dim.

• number of components: [1, 2, 3, 4, 5]
• lr: [0.005,0.001,0.0015,0.0005]
• embed_dim: [8, 16, 32, 64, 128, 256]