| 211 | 0 | 282 |
| 下载次数 | 被引频次 | 阅读次数 |
针对传统的时间序列预测方法在对现实场景的时间进行预测时精确度不高及效率低下等问题，提出一种改进的基于Mamba的时间序列预测算法。该算法以新型的具有线性时间复杂度的状态空间模型（SSM）为基础，在简单Mamba(S-Mamba)上进行改进。通过增加通道注意力与特征融合的方式建立CA-FF-Mamba模型。该模型在3个现实场景的公共数据集上进行验证，展现出良好的性能，均方误差（MSE）和平均绝对误差（MAE）均有显著提升。结果表明：采用通道加权与特征融合的方法，能够为时间序列预测研究提供更为高效的解决方案。
Abstract：To tackle the issues of low accuracy and inefficiency in traditional time series forecasting methods when applied to real-world scenarios, this paper proposes an improved Mamba-based algorithm for time series prediction. Building upon a novel state space models(SSM) with linear time complexity, the method improves on the Simple-Mamba(S-Mamba). A CAFF-Mamba(channel attention-feature fusion-mamba) model is established by adding channel attention and feature fusion.The proposed model is validated on public datasets of three real-world scenarios, demonstrating superior performance with significant reductions in both the mean squared error(MSE) and mean absolute error(MAE). The results indicate that the channel weighting and feature fusion approach offers a more efficient solution for time series forecasting research.
[1] HINTON G E,SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science,2006,313(5786):504-507.
[2] BAHDANAU D,CHO K,BENGIO Y. Neural machine translation by jointly learning to align and translate[EB/OL].(2014-09-01)[2024-03-02]. https：//arxiv.org/abs/1409. 0473.
[3] LECUN Y,BOTTOU L,BENGIO Y,et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE,1998,86(11):2278-2324.
[4] KRIZHEVSKY A,SUTSKEVER I,HINTON G E. ImageNet classification with deep convolutional neural networks[J].Communications of the ACM,2017,60(6):84-90.
[5] VASWANI A,SHAZEER N,PARMAR N,et al. Attention is All You Need[EB/OL].(2023-08-02)[2024-05-11]. https：//arxiv.org/abs/1706.03762.
[6] GU A,DAO T. Mamba:linear-time sequence modeling with selective state spaces[EB/OL].[2024-05-31]. https：//arxiv.org/abs/2312.00752
[7] ZENG C,LIU Z,ZHENG G,et al. C-Mamba:channel correlation enhanced state space models for multivariate time series forecasting[EB/OL].[2024-06-08]. https：//arxiv.org/abs/2406.05316.
[8] WU Z,GONG Y,ZHANG A. DTMamba:dual twin Mamba for time series forecasting[EB/OL].[2024-05-11]. https：//arxiv.org/abs/2405.07022.
[9] LI L C,WANG H C,ZHANG W J,et al. STG-mamba:spatial-temporal graph learning via selective state space model[EB/OL].[2024-03-19].https：//arxiv.org/abs/2403.12418.
[10] LIANG A,JIANG X,SUN Y,et al. Bi-Mamba+:bidirectional Mamba for time series forecasting[EB/OL].[2024-04-24]. https：//arxiv.org/abs/2404.15772.
[11]PATRO B N,AGNEESWARAN V. SiMBA:simplified mamba-based architecture for vision and multivariate time series[EB/OL].[2024-03-22]. https：//arxiv.org/abs/2403.15360.
[12] WANG Z,KONG F,FENG S,et al. Is Mamba effective for time series forecasting?[EB/OL].[2024-04-27]. https：//arxiv.org/abs/2403.11144.
[13] GU A,JOHNSON I,GOEL K,et al. Combining recurrent,convolutional,and continuous-time models with linear state space layers[J]. Advances in Neural Information Processing Systems,2021,34:572-585.
[14] GU A,GOEL K,RE C. Efficiently modeling long sequences with structured state spaces[EB/OL].(2022-08-05)[2024-04-23]. https：//arxiv.org/abs/2111.00396.
[15] GU A,DAO T,ERMON S,et al. HiPPO:recurrent memory with optimal polynomial projections[J]. Advances in Neural Information Processing Systems,2020,33:1474-1487.
[16] ZHANG H,ZHU Y,WANG D,et al. A survey on visual Mamba[J]. Applied Sciences,2024,14(13):5683.
[17] DAO T,FU D Y,SAAB K K,et al. Hungry hungry hippos:towards language modeling with state space models[EB/OL].(2023-04-29)[2024-03-19]. https：//arxiv.org/abs/2212.14052.
[18] HENDRYCKS D,GIMPEL K. Gaussian error linear units(GELUs)[EB/OL].(2023-06-06)[2024-04-24]. https：//arxiv.org/abs/1606.08415.
[19]RAMACHANDRAN P,ZOPH B,LE Q V. Searching for activation functions[EB/OL].(2017-10-27)[2024-05-11].https：//arxiv.org/abs/1710.05941.
[20]BA J,KIROS J,HINTON G E. Layer normalization[EB/OL].(2016-07-21)[2024-02-12]. https：//arxiv.org/abs/1607.06450.
[21]SUN Y T,DONG L,HUANG S H,et al. Retentive network:a successor to transformer for large language models[EB/OL].(2023-08-09)[2024-03-19]. https：//arxiv.org/abs/2307.08621.
[22]LIU Y,HU T G,ZHANG H R,et al. iTransformer:inverted transformers are effective for time series forecasting[EB/OL].[2024-03-14]. https：//arxiv.org/abs/2310.06625.
[23] HU J,SHEN L,SUN G. Squeeze-and-excitation networks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City,UT,USA:IEEE,2018:7132-7141.
[24] ZHOU T,MA Z Q,WEN Q S,et al. FEDformer:frequency enhanced decomposed transformer for long-term series forecasting[EB/OL].(2022-01-30)[2024-05-11]. https：//arxiv.org/abs/2201. 12740v3.
[25]WU H X,XU J H,WANG J M,et al. Autoformer:decomposition transformers with auto-correlation for long-term series forecasting[J]. Advances in Neural Information Processing Systems,2021,34:22419-22430.
[26]ZHANG Y,YAN J. Crossformer:Transformer utilizing crossdimension dependency for multivariate time series forecasting[C]//International Conference on Learning Representations(ICLR). Virtual Conference:OpenReview.net,2023:vSVLM2j9eie.
基本信息:
DOI：10.19573/j.issn2095-0926.202503002
中图分类号:O211.61;TP18
引用信息:
[1]翟宝英,王志强,刘璐,等.一种改进的S-Mamba时间序列预测算法[J].天津职业技术师范大学学报,2025,35(03):9-14.DOI:10.19573/j.issn2095-0926.202503002.
基金信息:
国家自然科学基金面上项目(11772222)