BERT-BiGRU Intelligent Classification of Metro On-Board Equipment Faults Based on Key Layer Fusion

Table 4

The classification effect of different classifiers on the data set.

Number	Method	Precision	Recall	-score
1	Fasttext	0.8147	0.8294	0.8219
2	CNN	0.8213	0.8272	0.8142
3	BiLSTM	0.8469	0.8267	0.8367
4	BiGRU	0.8454	0.8263	0.8357
5	2-BERT	0.8777	0.8859	0.8818
6	ALBERT-BiLSTM	0.9106	0.9041	0.9033
7	ALBERT-BiGRU	0.9064	0.9078	0.9071
8	2-BERT-BIGRU	0.9137	0.9087	0.9112
9	4-BERT-BiGRU	0.9109	0.9220	0.9164
10	8-BERT-BiGRU	0.9162	0.9265	0.9213
11	12-BERT-BiGRU	0.9203	0.9293	0.9248
12	The model proposed in paper	0.9231	0.9314	0.9272