Walk-forward empirical wavelet random vector functional link for time series forecasting

Gao, Ruobin; Du, Liang; Yuen, Kum Fai; Suganthan, Ponnuthurai Nagaratnam

doi:10.1016/j.asoc.2021.107450

Cited by 44 publications

(23 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Then, a k-fold cross-validation is conducted to tune the hyper-parameters [143]. The second approach splits the training and validation set in chronological order [105].…”

Section: Methodsmentioning

confidence: 99%

“…Group-wise tests can determine the overall ranking of the models on all the datasets and group them based on the statistical distance. The literature about RVFL utilizes the Nemenyi test to compare the models in a group-wise fashion [105,143]. The pair-wise tests assist in comparing the models in a pair-wise fashion, which is straightforward to show the better model.…”

Section: Methodsmentioning

confidence: 99%

“…Most literature utilizes a comprehensive grid search to tune these parameters. Grid search is straightforward to implement and succeeds on many tasks, such as classification [103,24], forecasting [105,169]. A comprehensive evaluation of RVFL for classification is conducted by Zhang and Suganthan [24].…”

Section: Hyper-parameters Optimization For Single-layer Rvflmentioning

confidence: 99%

“…Finally, tuning the scaling of randomization weights (input scaling) also increases the performance. Gao et al [105] divides the time series into training, validation, and test set in chronological order. Then, a grid search is conducted to select the hyperparameters, number of hidden nodes, and regularization strength, according to the forecasting errors on the validation set.…”

Section: Hyper-parameters Optimization For Single-layer Rvflmentioning

confidence: 99%

“…Accurate and reliable forecasts help the stakeholders and decision-makers plan, organize, maintain and develop the system in advance in a data-driven fashion. RVFL and the improved versions have demonstrated their outstanding performance on various forecasting tasks from different domains, such as electricity load [105,143], solar power [174], wind power [148], financial time series [133] and other data [108]. This section presents the details of all the literature about RVFL-based forecasting.…”

Section: Other Applications Of Rvfl Modelmentioning

confidence: 99%

See 4 more Smart Citations

Random vector functional link network: recent developments, applications, and future directions

Malik¹,

Gao²,

Ganaie³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Neural networks have been successfully employed in various domain such as classification, regression and clustering, etc. Generally, the back propagation (BP) based iterative approaches are used to train the neural networks, however, it results in the issues of local minima, sensitivity to learning rate and slow convergence. To overcome these issues, randomization based neural networks such as random vector functional link (RVFL) network have been proposed. RVFL model has several characteristics such as fast training speed, simple architecture, and universal approximation capability, that make it a viable randomized neural network. This article presents the comprehensive review of the development of RVFL model, which can serve as the extensive summary for the beginners as well as practitioners. We discuss the shallow RVFL, ensemble RVFL, deep RVFL and ensemble deep RVFL models. The variations, improvements and applications of RVFL models are discussed in detail. Moreover, we discuss the different hyperparameter optimization techniques followed in the literature to improve the generalization performance of the RVFL model. Finally, we give potential future research directions/opportunities that can inspire the researchers to improve the RVFL architecture further.

show abstract

“…Then, a k-fold cross-validation is conducted to tune the hyper-parameters [143]. The second approach splits the training and validation set in chronological order [105].…”

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%