Applying Deep Learning Technologies to Evaluate the Patent Quality with the Collaborative Training

Table 5

PQE-MT layer information.

Number of layers	Type	Output shape	Number of parameters
0	Text input layer	(None, 200)	0
1	Embedding layer	(None, 200, 768)	17999616
2	Bidirectional LSTM_1	(None, 200, 160)	543360
3	Bidirectional LSTM_2	(None, 200, 160)	154240
4	Attention	(None, 160)	32400
5	Number input layer	(None, 132)	0
6	Concatenate layer	(None, 292)	0
7/9/11	Dense layer_1/2/3	(None, 512/128/32)	150016
8/10/12	Dropout layer_1/2/3	(None, 512/128/32)	65664
13	CRF layer	(None, 200, 4)	4128
14	Softmax layer	(None, 8)	668