A Robust Text Classifier Based on Denoising Deep Neural Network in the Analysis of Big Data

Table 3

Text classification performance with different models using BBC news dataset.

	Classifier	Plus ratio noise
		0.00	0.01	0.02	0.03	0.04	0.05
Accuracy	NB	0.9659	0.9560	0.9339	0.8736	0.8186	0.7852
	KNN	0.9375	0.9325	0.9284	0.9373	0.9119	0.9260
	SVM	0.9715	0.9701	0.9672	0.9583	0.9340	0.9075
	DBN	0.9462	0.9434	0.9268	0.9076	0.8789	0.8479
	DDNN	0.9700	0.9685	0.9582	0.9541	0.9381	0.9286
Recall	NB	0.9655	0.9550	0.9294	0.8453	0.7387	0.6652
	KNN	0.9354	0.9324	0.9279	0.9369	0.9114	0.9249
	SVM	0.9715	0.9700	0.9670	0.9580	0.9309	0.8964
	DBN	0.9459	0.9429	0.9249	0.9039	0.8769	0.8393
	DDNN	0.9700	0.9685	0.9580	0.9535	0.9399	0.9249
-score	NB	0.9657	0.9555	0.9316	0.8592	0.7766	0.7202
	KNN	0.9364	0.9324	0.9281	0.9371	0.9116	0.9254
	SVM	0.9715	0.9700	0.9671	0.9581	0.9324	0.9019
	DBN	0.9460	0.9431	0.9258	0.9057	0.8779	0.8436
	DDNN	0.9700	0.9685	0.9581	0.9538	0.9390	0.9267