VulDistilBERT: A CPS Vulnerability Severity Prediction Method Based on Distillation Model

Table 11

Results of comparing the performance of different classifiers on the dataset integrated with the optimal subset.

Models	Classifiers	Best epoch	Loss	Accuracy	Precision	Recall	F1 scores
Pretrained DistilBERT	Linear	31	1.035	0.57	0.43	0.34	0.31
Fine-tuned DistilBERT	Linear	4	0.231	0.93	0.88	0.87	0.87
Fine-tuned DistilBERT	CNN	4	0.237	0.93	0.90	0.87	0.88
Fine-tuned DistilBERT	LSTM	4	0.234	0.93	0.86	0.90	0.88