Multitask Convolutional Neural Network for Rolling Element Bearing Fault Identification

Table 4

Experimental results of different modes.

		Joint training	Individual training
Train set	Classification accuracy	99.21%	99.37%
	Classification loss	0.0180	0.0103
	Regression loss	1.9414	10.566
Test set	Classification accuracy	99.11%	99.13%
	Classification loss	0.0324	0.0189
	Regression loss	2.4563	13.108