Research Article
Exploiting the Relationship between Pruning Ratio and Compression Effect for Neural Network Model Based on TensorFlow
Table 5
Descriptions of parameters of the pruning effect in the second scene.
| ā | Original number of parameters | Threshold value | Valid parameter number |
| w_conv1 | 800 | 0.0 | 800 | w_conv2 | 51,200 | 0.0 | 51,200 | w_fc1 | 3,211,264 | 0.036936015 | 2,247,884 | w_fc2 | 10,240 | 0.039559085 | 7,167 |
|
|