Please use this identifier to cite or link to this item:
Title: Learning Nonstationary Time-Series With Dynamic Pattern Extractions
Authors: Wang, X
Zhang, H
Zhang, Y
Wang, M
Song, J
Lai, T
Khushi, M
Keywords: attention;forecast;gated recurrent units (GRU);recurrent neural network (RNN);Seq2seq;time-series
Issue Date: 24-Nov-2021
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Citation: Wang, X. et al. (2021) 'Learning Nonstationary Time-Series With Dynamic Pattern Extractions', IEEE Transactions on Artificial Intelligence, 3 (5), pp. 778 - 787. doi: 10.1109/TAI.2021.3130529.
Abstract: Copyright © The authors 2021. The era of information explosion had prompted the accumulation of a tremendous amount of time-series data, including stationary and non-stationary time-series data. State-of-the-art algorithms have achieved a decent performance in dealing with stationary temporal data. However, traditional algorithms that tackle stationary time-series do not apply to non-stationary series like Forex trading. This paper investigates applicable models that can improve the accuracy of forecasting future trends of non-stationary time-series sequences. In particular, we focus on identifying potential models and investigate the effects of recognizing patterns from historical data. We propose a combination of \rebuttal{the} seq2seq model based on RNN, along with an attention mechanism and an enriched set features extracted via dynamic time warping and zigzag peak valley indicators. Customized loss functions and evaluating metrics have been designed to focus more on the predicting sequence's peaks and valley points. Our results show that our model can predict 4-hour future trends with high accuracy in the Forex dataset, which is crucial in realistic scenarios to assist foreign exchange trading decision making. We further provide evaluations of the effects of various loss functions, evaluation metrics, model variants, and components on model performance.
Other Identifiers: ORCID iDs: Xipei Wang; Meng Wang; Jiarui Song; Tin Lai; Matloob Khushi
Appears in Collections:Dept of Computer Science Embargoed Research Papers

Files in This Item:
File Description SizeFormat 
FullText.pdfCopyright © The authors 2021. Archived under a Creative Commons (CC BY) Creative Commons ( on arXiv at arXiv:2111.10559v1 [cs.LG] for this version). (see: The version of record is available at, copyright © 2021 Institute of Electrical and Electronics Engineers (IEEE). Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works by sending a request to For more information, see MBAdobe PDFView/Open

This item is licensed under a Creative Commons License Creative Commons