基于多注意力的多变量时间序列特征选择方法  

A Multi-attention-based Feature Selection Method for Multivariate Time Series

作　　者：胡紫音 桂宁[2] HU Zi-yin;GUI Ning(School of Information Science and Technology,Zhejiang Sci-Tech University,Hangzhou 310018,China;School of Computer Science,Central South University,Changsha 410006,China)

机构地区：[1]浙江理工大学信息学院,浙江杭州310018 [2]中南大学计算机学院,湖南长沙410006

出　　处：《软件导刊》2020年第11期21-24,共4页Software Guide

基　　金：国家自然科学基金项目(61772473)。

摘　　要：特征选择是避免维度诅咒的一种数据预处理技术。在多变量时间序列预测中,为了同时找到与问题相关性最大的变量及其对应时延,提出一种基于多注意力的有监督特征选择方法。该方法利用带有注意力模块和学习模块的深度学习模型,将原始二维时间序列数据正交分割成两组一维数据,分别输入两个不同维度的注意力生成模块,得到特征维度和时间维度的注意权重。两个维度的注意力权值点积叠加作为全局注意力得分进行特征选择,作用于原始数据后输入随学习模块训练不断更新至收敛。实验结果表明,所提出的方法在特征数小于10时可达到全量数据训练效果,与现有几种基线方法相比实现了最佳准确率。Feature selection is a data preprocessing technique that reduces model complexity and avoids the curse of dimensionality.In order to find the variable that is most relevant to the problem and its corresponding delay simultaneously in multivariate time series pre⁃diction,this paper proposes a multi-attention based supervised feature selection method.This method uses a deep learning model with an attention module and a learning module.The original two-dimensional time series data is orthogonally divided into two sets of onedimensional data and input into the attention module of two different dimensions respectively to generate the attention weights of the fea⁃ture dimension and the time dimension.Then the attention weights of the two dimensions are dotted with the product operation,used as a global attention score for feature selection,applied to the original data and updated continuously with the training process until the model converges.Experimental results show that the proposed method can achieve the effect of full data training when the number of features is less than 10,and achieves the best accuracy compared with several existing baseline methods.

关 键 词：特征选择 时间序列 注意力机制 多维数据 深度学习 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]