Fault Identification and Analysis of Communication Network Based on Deep Learning

Table 10

Comparison of accuracy of fault diagnosis models before and after feature screening after training stabilization.

Accuracy	Algorithm
	Bayesian (%)	KNN (%)	Decision tree (%)	Forest stochastic (%)	CNN (%)	XGBoost (%)
After feature screening	92.37	97.11	98.87	98.48	97.33	98.92
Before feature screening	89.01	87.44	88.10	88.22	84.31	91.26