دورية أكاديمية
A Joint Time-Frequency Domain Transformer for multivariate time series forecasting.
| العنوان: | A Joint Time-Frequency Domain Transformer for multivariate time series forecasting. |
|---|---|
| المؤلفون: | Chen Y; Department of Computer Science and Technology, Tsinghua University, RM.3-126, FIT Building, Haidian District, Beijing, 100084, China., Liu S; College of Computer Science and Mathematics, Fujian University of Technology, RM.213, Building C4, Fuzhou, Fujian, 350118, China., Yang J; Techorigin, No. 581, Jianzhu West Road, Binhu District, Wuxi, Jiangsu, 214000, China., Jing H; Earth System Modeling and Prediction Center, No. 46, Zhongguancun South Street, Haidian District, Beijing, 100081, China., Zhao W; Department of Computer Science and Technology, Tsinghua University, RM.3-126, FIT Building, Haidian District, Beijing, 100084, China., Yang G; Department of Computer Science and Technology, Tsinghua University, RM.3-126, FIT Building, Haidian District, Beijing, 100084, China. Electronic address: ygw@tsinghua.edu.cn. |
| المصدر: | Neural networks : the official journal of the International Neural Network Society [Neural Netw] 2024 Aug; Vol. 176, pp. 106334. Date of Electronic Publication: 2024 Apr 25. |
| نوع المنشور: | Journal Article |
| اللغة: | English |
| بيانات الدورية: | Publisher: Pergamon Press Country of Publication: United States NLM ID: 8805018 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1879-2782 (Electronic) Linking ISSN: 08936080 NLM ISO Abbreviation: Neural Netw Subsets: MEDLINE |
| أسماء مطبوعة: | Original Publication: New York : Pergamon Press, [c1988- |
| مواضيع طبية MeSH: | Forecasting*, Time Factors ; Neural Networks, Computer ; Algorithms ; Multivariate Analysis ; Humans |
| مستخلص: | In order to enhance the performance of Transformer models for long-term multivariate forecasting while minimizing computational demands, this paper introduces the Joint Time-Frequency Domain Transformer (JTFT). JTFT combines time and frequency domain representations to make predictions. The frequency domain representation efficiently extracts multi-scale dependencies while maintaining sparsity by utilizing a small number of learnable frequencies. Simultaneously, the time domain (TD) representation is derived from a fixed number of the most recent data points, strengthening the modeling of local relationships and mitigating the effects of non-stationarity. Importantly, the length of the representation remains independent of the input sequence length, enabling JTFT to achieve linear computational complexity. Furthermore, a low-rank attention layer is proposed to efficiently capture cross-dimensional dependencies, thus preventing performance degradation resulting from the entanglement of temporal and channel-wise modeling. Experimental results on eight real-world datasets demonstrate that JTFT outperforms state-of-the-art baselines in predictive performance. Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. (Copyright © 2024 Elsevier Ltd. All rights reserved.) |
| فهرسة مساهمة: | Keywords: Frequency domain; Multivariate; Time series forecasting; Transformer |
| تواريخ الأحداث: | Date Created: 20240430 Date Completed: 20240615 Latest Revision: 20240615 |
| رمز التحديث: | 20240616 |
| DOI: | 10.1016/j.neunet.2024.106334 |
| PMID: | 38688070 |
| قاعدة البيانات: | MEDLINE |
Full Text Finder | تدمد: | 1879-2782 |
|---|---|
| DOI: | 10.1016/j.neunet.2024.106334 |