“…Various machine learning approaches such as neural networks, support vector machines, Naïve Bayes classifier, etc., have been used for fault diagnosis under the broad umbrella of data-driven methods. Neural-network-based fault-diagnosis approaches [13,14] have included, for feature generation: kurtosis and entropy [15], wavelet transforms [16], and fractional wavelet transforms [17]; and for dimensionality reduction: kernel PCA (kPCA) [16,17]. Support vector machine (SVM)-based [18] fault-diagnosis approaches have further included, for feature generation: fractional Fourier transform [19], cross-wavelet transform [20,21], deep belief networks (DBN) [22,23], and empirical mode decomposition [24]; for dimensionality reduction: parametric t-SNE [20] and principal component analysis [21]; and for SVM hyperparameter optimization: the double-chains quantum genetic algorithm [24], the fruitfly algorithm [25], the barnacles mating optimizer algorithm [26], and the firefly algorithm [27].…”