In current time series prediction tasks, many methods transform time series into the frequency domain for analysis. Analyzing time series in the frequency domain reveals the spectral characteristics and energy distribution across different frequencies, providing a clearer understanding of the contributions of various frequency components. However, existing methods using frequency domain analysis often overlook the characteristic information contained within the amplitude and phase in the same frequency range. Additionally, many frequency domain analysis models paired with transformers result in significant time complexity. Therefore, we designed a parallel dual-channel network that processes frequencies categorically and then decomposes the complex values in the frequency domain to obtain amplitude and phase information within the same frequency range. This allows for capturing the intensity and relative position of corresponding frequency components in the entire time series. To capture the variations of amplitude and phase within the same frequency range in the time series, a frequency domain complex decomposition module was designed for refined feature extraction. Finally, the extracted amplitude-phase features are recombined into a complex number, providing a more accurate description of the distribution characteristics of frequency components in the time series. This method helps the model better understand frequency variations within the time series, thus improving prediction accuracy. Experiments show that, compared to the current state-of-the-art models, the proposed model achieves an average mean squared error (MSE) reduction of 6.29% to 18.90% across seven datasets including ETT and Weather, with significantly lower temporal and spatial complexity.
ZhangFWangMZhangW, et al.THATSN: temporal hierarchical aggregation tree structure network for long-term time-series forecasting. Inf Sci (Ny)2025; 692: 121659.
2.
HuangSLiuYZhangF, et al.CrossWaveNet: a dual-channel network with deep cross-decomposition for long-term time series forecasting. Expert Syst Appl2024; 238: 121642.
3.
ZhouXPanZHuG, et al.Stock market prediction on high-frequency data using generative adversarial nets. Math Probl Eng2018; 2018: 4907423.
4.
LiuXGuoJWangH, et al.Prediction of stock market index based on ISSA-BP neural network. Expert Syst Appl2022; 204: 117604.
5.
PangXZhouYWangP, et al.An innovative neural network approach for stock market prediction. J Supercomput2020; 76: 2098–2118.
6.
BiKXieLZhangH, et al.Accurate medium-range global weather forecasting with 3D neural networks. Nature2023; 619: 533–538.
ZhongJWangY. Enhancing thyroid disease prediction using machine learning: a comparative study of ensemble models and class balancing techniques, 2025.
9.
WangYZhongJKumarR. A systematic review of machine learning applications in infectious disease prediction, diagnosis, and outbreak forecasting, 2025.
10.
YuanLWangHZhangF. ADMNet: an adaptive downsampling multi-frequency multi-channel network for long-term time series forecasting. Expert Syst Appl2025; 262: 125588.
11.
ZhangWWangHZhangF. Skip-timeformer: skip-time interaction transformer for long sequence time-series forecasting. In: International joint conference on artificial intelligence, 2024, pp.5499–5507.
12.
LeeY-STongL-I. Forecasting time series using a methodology based on autoregressive integrated moving average and genetic programming. Knowl Based Syst2011; 24: 66–72.
SmylS. A hybrid method of exponential smoothing and recurrent neural networks for time series forecasting. Int J Forecast2020; 36: 75–85.
15.
Siami-NaminiSTavakoliNNaminAS. A comparison of ARIMA and LSTM in forecasting time series. In: 2018 17th IEEE international conference on machine learning and applications (ICMLA), IEEE, 2018, pp.1394–1401.
16.
TaoHLiJHuaZ, et al.DUDB: deep unfolding based dual-branch feature fusion network for pan-sharpening remote sensing images. IEEE Trans Geosci Remote Sens2023; 62: 1–17.
17.
ZhangFChenGWangH, et al.CF-DAN: facial-expression recognition based on cross-fusion dual-attention network. Comput Visual Media2024; 10: 593–608.
18.
ZhangFWangMLiL, et al.Probabilistic intervals prediction based on adaptive regression with attention residual connections and covariance constraints. Eng Appl Artif Intell2025; 156: 111013.
19.
Al FahoumAZyoutA. Wavelet transform, reconstructed phase space, and deep learning neural networks for EEG-based schizophrenia detection. Int J Neural Syst2024; 34: 2450046.
20.
Al FahoumAAl OmariAAl OmariG, et al.Development of a novel light-sensitive PPG model using PPG scalograms and PPG-NET learning for non-invasive hypertension monitoring. Heliyon2024; 10: e39745.
21.
Al FahoumA. Complex wavelet-enhanced convolutional neural networks for electrocardiogram-based detection of paroxysmal atrial fibrillation. Adv Signal Process Artif Intell2024; 158: 2450046.
22.
WangYZhaoTWangX. Fine-grained heartbeat waveform monitoring with RFID: a latent diffusion model. In: Proceedings of the 3rd international workshop on human-centered sensing, modeling, and intelligent systems, 2025, pp.86–91.
23.
LiuMZengAChenM, et al.SCINet: time series modeling and forecasting with sample convolution and interaction. Adv Neural Inf Process Syst2022; 35: 5816–5828.
24.
ZhangYYanJ. Crossformer: transformer utilizing cross-dimension dependency for multivariate time series forecasting. In: The eleventh international conference on learning representations, 2022.
25.
ZhouTMaZWenQ, et al.Fedformer: frequency enhanced decomposed transformer for long-term series forecasting. In: International conference on machine learning, PMLR, 2022, pp.27268–27286.
26.
WuHHuTLiuY, et al.Timesnet: temporal 2d-variation modeling for general time series analysis. In: The eleventh international conference on learning representations, 2022.
27.
LinFGaoCYamadaKD. An effective convolutional neural network for visualized understanding transboundary air pollution based on Himawari-8 satellite images. IEEE Geosci Remote Sens Lett2021; 19: 1–5.
28.
YamadaKDLinFNakamuraT. Developing a novel recurrent neural network architecture with fewer parameters and good learning performance. Interdiscip Inf Sci2021; 27: 25–40.
29.
LinFXuYZhangZ, et al.Cosmos propagation network: deep learning model for point cloud completion. Neurocomputing2022; 507: 221–234.
30.
SuYLanTWangY, et al.A contrastive framework for neural text generation. Adv Neural Inf Process Syst2022; 35: 21548–21561.
31.
ZhongSScarinciACicirelloA. Natural language processing for systems engineering: automatic generation of systems modelling language diagrams. Knowl Based Syst2023; 259: 110071.
32.
ZhangFChenGWangH, et al.Multi-scale video super-resolution transformer with polynomial approximation. IEEE Trans Circuits Syst Video Technol2023; 33: 4496–4506.
33.
NguyenMKAyadILarueN. QN-Mixer: a quasi-newton mlp-mixer model for sparse-view CT reconstruction. In: IEEE/CVF computer vision and pattern recognition (CVPR), 2024.
34.
ChenGWangHLiuY, et al.Resformer: combine quadratic linear transformation with efficient sparse transformer for long-term series forecasting. Intell Data Anal2023; 27: 1557–1572.
35.
ZhangFGuoTWangH. DFNet: decomposition fusion model for long sequence time-series forecasting. Knowl Based Syst2023; 277: 110794.
36.
TokgözAÜnalG. A RNN based time series approach for forecasting turkish electricity load. In: 2018 26th Signal processing and communications applications conference (SIU), IEEE, 2018, pp.1–4.
37.
LivierisIEPintelasEPintelasP. A CNN–LSTM model for gold price time-series forecasting. Neural Comput Appl2020; 32: 17351–17360.
38.
YamakPTYujianLGadoseyPK. A comparison between ARIMA, LSTM, and GRU for time series forecasting. In: Proceedings of the 2019 2nd international conference on algorithms, computing and artificial intelligence, 2019, pp.49–55.
39.
SalinasDFlunkertVGasthausJ, et al.DeepAR: probabilistic forecasting with autoregressive recurrent networks. Int J Forecast2020; 36: 1181–1191.
40.
CaiWLiangYLiuX, et al.MSGNet: Learning multi-scale inter-series correlations for multivariate time series forecasting. In: Proceedings of the AAAI conference on artificial intelligence, Vol. 38, 2024, pp.11141–11149.
41.
LuoDWangX. ModernTCN: a modern pure convolution structure for general time series analysis. In: The twelfth international conference on learning representations, 2024.
42.
XuYAraiSLiuD, et al.FPCC: fast point cloud clustering-based instance segmentation for industrial bin-picking. Neurocomputing2022; 494: 255–268.
43.
LinFYueYHouS, et al.Hyperbolic chamfer distance for point cloud completion. In: Proceedings of the IEEE/CVF international conference on computer vision, 2023, pp.14595–14606.
44.
UzunISAmiraABouridaneA. FPGA implementations of fast Fourier transforms for real-time signal and image processing. IEE Proc Vision Image Signal Process2005; 152: 283–296.
45.
EleftheriadisCKarakonstantisG. Energy-efficient fast Fourier transform for real-valued applications. IEEE Trans Circuits Syst II Express Briefs2022; 69: 2458–2462.
46.
ChiLJiangBMuY. Fast Fourier convolution. Adv Neural Inf Process Syst2020; 33: 4479–4488.
47.
DuhamelPVetterliM. Fast Fourier transforms: a tutorial review and a state of the art. Signal Process1990; 19: 259–299.
48.
AhmedNNatarajanTRaoKR. Discrete cosine transform. IEEE Trans Comput1974; 100: 90–93.
49.
XuZZengAXuQ. FITS: modeling time series with 10 k parameters. arXiv preprint arXiv:2307.03756, 2023.
50.
NieYNguyenNHSinthongP, et al.A time series is worth 64 words: long-term forecasting with transformers. arXiv preprint arXiv:2211.14730, 2022.
51.
ZengAChenMZhangL, et al.Are transformers effective for time series forecasting? In: Proceedings of the AAAI conference on artificial intelligence, Vol. 37, 2023, pp.11121–11128.
52.
WangHPengJHuangF, et al.MICN: multi-scale local and global context modeling for long-term series forecasting. In: The eleventh international conference on learning representations, 2022.
53.
VaswaniAShazeerNParmarN, et al.Attention is all you need. Adv Neural Inf Process Syst2017; 30: 5988.
