Modes decomposition method in fusion with robust random vector functional link network for crude oil price forecasting

Carbon price prediction for China's ETS pilots using variational mode decomposition and optimized extreme learning machine

“…Due to the complexity of carbon price features, VMD algorithm could process original carbon price series into multiple IMFs with relatively stationarity and different frequencies, resulting in improved forecasting precision. However, there is no adaptive processing procedure or clear standard to confirm the best number of modes so far (Bisoi et al, 2019a(Bisoi et al, , 2019bJiang et al, 2019). Existing studies have determined the parameter by selecting the number of IMFs generated from EMD or EEMD algorithm (Bisoi et al, 2019a(Bisoi et al, , 2019bZhu et al, 2019).…”

“…However, there is no adaptive processing procedure or clear standard to confirm the best number of modes so far (Bisoi et al, 2019a(Bisoi et al, , 2019bJiang et al, 2019). Existing studies have determined the parameter by selecting the number of IMFs generated from EMD or EEMD algorithm (Bisoi et al, 2019a(Bisoi et al, , 2019bZhu et al, 2019). Tian and Hao (2020) proposed that the parameter settings of EEMD may influence decomposition results, whereas EMD demands no specific parameter to be set.…”

An Improved GWO Algorithm Optimized RVFL Model for Oil Layer Prediction

“…The number of nodes is increased from 10 to 150, the activa function of each improved model algorithm is set to sigmoid, the dataset is divided five folds for cross-validation, and it is run 10 times to take the average as the final re The experimental results are shown in Table A3 and Figure 5, which visually show trend of prediction accuracy of various improved RVFL algorithms with the numb nodes. As seen in Table A3, when the number of nodes is in the interval [10,110], the ave prediction accuracy of each algorithm improves significantly as the number of nod the hidden layer increases. As seen in Figure 6, the average prediction accuracy of As seen in Table A3, when the number of nodes is in the interval [10,110], the average prediction accuracy of each algorithm improves significantly as the number of nodes in the hidden layer increases.…”

“…Tang et al [9] proposed an ensemble empirical mode decomposition (EEMD) technique for the RVFL model to improve prediction accuracy. Bisoi et al [10] combined variational mode decomposition (VMD) with an RVFL neural network model to improve both the running time and prediction accuracy of the code. Yu et al [11] synthesized the impact of five different strategies in the predictive performance of RVFL neural network models from the perspective of the diversity of integration strategies.…”

Multi‐kernel‐based random vector functional link network with decomposed features for epileptic EEG signal classification

“…Also, backpropagation (BP) algorithm is applied in the soft-max layer to reach the final solution for classification that might exhibit lesser ACC due to slower convergence, local minima, and over fitting problems. To overcome such issues several non-iterative randomised neural networks such as extreme learning machine (ELM) [36][37][38][39][40][41][42] and random vector functional link networks (RVFLN) [43,44] have witnessed several applications in regression and classification problems in recent years since they use a simple closed form solution by optimising a squared loss function with l2 regularisation by the least-squares approach. Randomisation-based neural networks greatly benefit from the presence of direct links from the input layer to the output layer as in RVFLN.…”