Table 5 The parameters of knowledge completion model
From: Using knowledge graphs and deep learning algorithms to enhance digital cultural heritage management
Category | Parameters | Rate |
|---|---|---|
Attention layer | Learning rate | 1e−3 |
Weight decay | 5e−6 | |
Epoch | 3600 | |
Negative ratio | 2 | |
Dropout | 0.3 | |
Leaky ReLU | 0.2 | |
Filters | None | |
ConvKB layer | Learning rate | 1e−5 |
Weight decay | 1e−5 | |
Epoch | 200 | |
Negative ratio | 40 | |
Dropout | 0.0 | |
Leaky ReLU | None | |
Filters | 500 |