PBDT: Python Backdoor Detection Model Based on Combined Features

Table 7

Performance comparison with other models.

	Accuracy	Precision	Recall (TPR)	TNR	F1-score
Guo et al. [38]	0.8355	0.6393	0.8696	0.8233	0.7368
Cui [18]	0.9122	0.8961	0.7565	0.9682	0.8201
Fang et al. [22]	0.9391	0.8563	0.9255	0.9441	0.8896
PBDT	0.9770	0.9623	0.9503	0.9866	0.9563