Fault Diagnosis for Rotating Machinery Based on Convolutional Neural Network and Empirical Mode Decomposition

Table 3

Training accuracy of both classifiers on different features.

Features	SVM	Softmax
11 time-domain and EMD features	88.72%; (36908/41600)	89.47%; (37220/41600)
80 CNN features	100%	100%
91 combined features	100%	100%