A Graph Convolutional Network-Based Sensitive Information Detection Algorithm

Table 3

Parameters for the proposed algorithm.

---

 The pretrained BERT (700d) is chosen as the fixed word embedding

 Maximal size of neighborhood (connected nodes) is 5

 The dropout rate to 0.5

 The learning rate to 0.0008 with the Adam optimizer

 Maximum number of training iterations is 1000

---