“…In order to cater for various application scenarios, many variants of ELM have been proposed, such as convex ELM [7], ordinal ELM [8], semisupervised ELM [9], Bayesian ELM [10], voting based ELM [11], PCA-ELM [12], robust ELM [13], microgenetic ELM [14], fuzzy ELM [15], hierarchical ELM [16], kernel ELM [17], ELM for multilayer perceptron [18], stacked ELM [19], wavelet ELM [20], graph embedded ELM [21], ELM for interval neural networks [22], memetic ELM [23], and ELM with binary output layer [24], ELM for imbalance learning [25], and ELM for residual learning [26]. The readers can refer to [27]- [29] for a thorough review of the essence and trends of ELM. Recently online sequential ELM has been applied to various fields such as fault diagnosis [30] and air quality forecasting [31].…”