Online support vector quantile regression for the dynamic time series with heavy-tailed noise

In this paper, we propose an online support vector quantile regression approach with an s-insensitive pinball loss function, called Online-SVQR, for dynamic time series with heavy-tailed noise. Online-SVQR is robust to heavy-tailed noise, as it can control the negative influence of heavy-tailed noise by using a quantile parameter. By using an incremental learning algorithm to update the new samples, the coefficients of Online-SVQR reflect the dynamic information in the examined time series. During each incremental training process, the nonsupport vector is ignored while the support vector continues training with new updated samples. Online-SVQR can select useful training samples and discard irrelevant samples. As a result, the training speed of Online-SVQR is accelerated. Experimental results on one artificial dataset and three real-world datasets indicate that Online-SVQR outperforms epsilon-support vector quantile regression in terms of both sample selection ability and training speed. (C) 2021 Elsevier B.V. All rights reserved.

Automated summary

The paper introduces an online support vector quantile regression method, Online-SVQR, for dynamic time series with heavy-tailed noise. By using an incremental learning algorithm to update coefficients, Online-SVQR reflects dynamic information and outperforms traditional epsilon-support vector quantile regression in terms of sample selection ability and training speed.