融合双通路注意力与VT-LSTM的金融时序预测

doi:10.3778/j.issn.1002-8331.2204-0066

摘要/Abstract

摘要： 针对现有研究对金融时序数据短期变化规律捕捉能力不足和预测精度不佳的问题，提出一种基于双通路注意力机制和改进转换门控LSTM（variant transformation-gated LSTM，VT-LSTM）的金融时序预测模型（dual-attention MDWT-CVT-LSTM）。使用多级离散小波变换（MDWT）分解股指序列得到高频和低频数据，并在融合门控单元的LSTM中加入转换门控机制，构造VT-LSTM，其能有效把控短期突变信息。在双通路注意力网络中结合VT-LSTM与一维时序卷积（Conv1D），分别提取不同频度数据的空间局部特征和时序特征，对各子序列进行预测，实现多层级多通路的预测研究。在金融股指数据集和个股数据集上对不同模型进行实验比较，结果表明提出模型预测精度优于其他方法，有良好的可行性。

关键词: 金融时间序列, 双通路注意力机制, 时序卷积, 多级离散小波变换, 长短时记忆网络

Abstract: To address the problems of existing research on the inadequate ability to capture short-term change patterns of financial time series data and poor prediction accuracy, a financial time series prediction model（dual-attention MDWT-CVT-LSTM） based on dual-pathway attention mechanism and improved transformation-gated LSTM（VT-LSTM） is proposed. Firstly, this paper decomposes the stock index series using multi-level discrete wavelet transform（MDWT） to obtain high and low frequency data, and adds the transformation-gated mechanism to the LSTM of fusion gating unit to construct the VT-LSTM, which can effectively grasp short-term mutation information. Then, VT-LSTM is combined with one-dimensional temporal convolution（Conv1D） in a dual-pathway attention network to extract spatial local features and temporal features of different frequency data respectively, and finally, each sub-series is predicted to realize a multi-level and multi-pathway prediction study. The experimental comparison of different models on financial stock market index datasets and individual stocks datasets shows that the proposed model has better prediction accuracy than other methods and good feasibility.

Key words: financial time series, dual-attention mechanism, time-series convolution, multilevel discrete wavelet transform, long and short time memory networks

戴宇睿, 安俊秀, 陶全桧. 融合双通路注意力与VT-LSTM的金融时序预测[J]. 计算机工程与应用, 2023, 59(12): 157-165.

DAI Yurui, AN Junxiu, TAO Quanhui. Financial Time-Series Prediction by Fusing Dual-Pathway Attention with VT-LSTM[J]. Computer Engineering and Applications, 2023, 59(12): 157-165.

参考文献

[1] RANGEL-GONZALEZ J A，FRAUSTO-SOLIS J，GONZáLEZ-
BARBOSA J J，et al.Comparative study of ARIMA methods for forecasting time series of the mexican stock exchange[J].Studies in Computational Intelligence，2018，749：475-485.
[2] CHEN Y，HAO Y.A feature weighted support vector machine and K-nearest neighbor algorithm for stock market indices prediction[J].Expert Systems with Applications，2017，80：340-355.
[3] 孙笑笑，侯文杰，应钰柯，等.基于双层机器学习的业务流程剩余时间预测[J].计算机学报，2021，44（11）：2283-2294.
SUN X X，HOU W J，YING Y K，et al.Business process remaining time prediction based on two-layer machine learning[J].Chinese Journal of Computers，2021，44（11）：2283-2294.
[4] PRIETO A，PRIETO B，ORTIGOSA E M，et al.Neural networks：an overview of early research，current frameworks and new challenges[J].Neurocomputing，2016，214：242-268.
[5] CHONG E，HAN C，PARK F C.Deep learning networks for stock market analysis and prediction：methodology，data representations，and case studies[J].Expert Systems with Application，2017，83：187-205.
[6] 程艳，尧磊波，张光河，等.基于注意力机制的多通道CNN和BiGRU的文本情感倾向性分析[J].计算机研究与发展，2020，57（12）：2583-2595.
CHENG Y，YAO L B，ZHANG G H，et al.Text sentiment orientation analysis of multi-channels CNN and BiGRU based on attention mechanism[J].Journal of Computer Research and Development，2020，57（12）：2583-2595.
[7] 李卫疆，漆芳，余正涛.基于多通道特征和自注意力的情感分类方法[J].软件学报，2021，32（9）：2783-2800.
LI W J，QI F，YU Z T.Sentiment classification method based on multi-channel features and self-attention[J].Journal of Software，2021，32（9）：2783-2800.
[8] 赵红蕊，薛雷.基于LSTM-CNN-CBAM模型的股票预测研究[J].计算机工程与应用，2021，57（3）：203-207.
ZHAO H R，XUE L.Research on stock forecasting based on LSTM-CNN-CBAM model[J].Computer Engineering and Applications，2021，57（3）：203-207.
[9] LU W，LI J，WANG J，et al.A CNN-BiLSTM-AM method for stock price prediction[J].Neural Computing and Applications，2020：1-13.
[10] 李春，高飞，王会青.改进果蝇算法优化CIAO-LSTM网络的时序预测模型[J].计算机工程与应用，2020，56（11）：129-134.
LI C，GAO F，WANG H Q.Improved fruit fly optimization algorithm for optimizing time series prediction model of CIAO-LSTM network[J].Computer Engineering and Applications，2020，56（11）：129-134.
[11] 方义秋，卢壮，葛君伟.联合RMSE损失LSTM-CNN模型的股价预测[J].计算机工程与应用，2022，58（9）：294-302.
FANG Y Q，LU Z，GE J W.Forecasting stock prices with combined RMSE loss LSTM-CNN model[J].Computer Engineering and Applications，2022，58（9）：294-302.
[12] HU J，ZHENG W.Transformation-gated LSTM：efficient capture of short-term mutation dependencies for multivariate time series prediction tasks[C]//The International Joint Conference on Neural Networks，2019.
[13] LIU Y，GONG C，YANG L，et al.DSTP-RNN：a dual-stage two-phase attention-based recurrent neural networks for long-term and multivariate time series prediction[J].Expert Systems with Applications，2019.
[14] XIAO Y，YIN H，ZHANG Y，et al.A dual‐stage attention‐based Conv‐LSTM network for spatio‐temporal correlation and multivariate time series prediction[J].International Journal of Intelligent Systems，2021（11）.
[15] 李梦莹，王晓东，阮书岚，等.基于双路注意力机制的学生成绩预测模型[J].计算机研究与发展，2020，57（8）：1729-1740.
LI M Y，WANG X D，RUAN S L，et al.Student performance prediction model based on two-way attention mechanism[J].Journal of Computer Research and Development，2020，57（8）：1729-1740.
[16] QIAO W，YANG Z.Forecast the electricity price of U.S.using a wavelet transform-based hybrid model[J].Energy，2020：116704.
[17] SHI G，QIN C，TAO J，et al.A VMD-EWT-LSTM-based multi-step prediction approach for shield tunneling machine cutterhead torque[J].Knowledge-Based Systems，2021，228：107213.
[18] 易靖韬，严欢.基于小波分解和ARIMA-GRU混合模型的外贸风险预测预警研究[J/OL].中国管理科学：1-11[2022-03-27].DOI：10.16381/j.cnki.issn1003-207x.2021.1174.
YI J T，YAN H.Early prediction and warning of international trade risks based on wavelet decompositionand ARIMA-GRU hybrid model[J].Chinese Journal of Management Science：1-11[2022-03-27].DOI：10.16381/j.cnki.issn1003-207x.2021.1174.
[19] ZHANG Y，CHEN B，PAN G，et al.A novel hybrid model based on VMD-WT and PCA-BP-RBF neural network for short-term wind speed forecasting[J].Energy Conversion and Management，2019，195：180-197.
[20] WANG J，WANG Z，LI J，et al.Multilevel wavelet decomposition network for interpretable time series analysis[C]//Proceedings of the 24th ACM International Conference on Knowledge Discovery and Data Mining（KDD’18），2018：2437-2446.
[21] 万晨.基于深度学习的多变量时间序列预测算法与应用研究[D].南京：南京大学，2020.
WAN C.Research on deep learning based multi-varivate time series forecasting algorithms and application[D].Nanjing：Nanjing University，2020.
[22] 王艺霏，于雷，滕飞，等.基于长-短时序特征融合的资源负载预测模型[J].计算机应用，2022，42（5）：1508-1515.
WANG Y F，YU L，TENG F，et al.Resource-load prediction model based on long-short time series feature fusion[J].Journal of Computer Applications，2022，42（5）：1508-1515.
[23] HU J，ZHENG W.Multistage attention network for multivariate time series prediction[J].Neurocomputing，2020，383：122-137.
[24] HU J，LI B.A deep learning framework based on spatio-temporal attention mechanism for traffic prediction[C]//2020 IEEE 22nd International Conference on High Performance Computing and Communications；IEEE 18th International Conference on Smart City；IEEE 6th International Conference on Data Science and Systems（HPCC/SmartCity/DSS），2020.