首页> 外国专利> Computing devices, computer implementations, and computer-readable storage media for multivariate time series modeling and prediction (stochastic non-linear relationships and external factors across multiple time series for improved multivariate time series modeling and prediction)

Computing devices, computer implementations, and computer-readable storage media for multivariate time series modeling and prediction (stochastic non-linear relationships and external factors across multiple time series for improved multivariate time series modeling and prediction)

机译:用于多元时间序列建模和预测的计算设备、计算机实现和计算机可读存储介质(用于改进多元时间序列建模和预测的跨多个时间序列的随机非线性关系和外部因素)

摘要

PROBLEM TO BE SOLVED: To provide a computing device, a computer mounting method and a computer readable storage medium for multivariate time series modeling and prediction. A computer-implemented method of time-series modeling and prediction 600 encodes a plurality of inputs of time-series data 605, maps the encoded multivariate time-series data to a low-dimensional latent space, and maps the encoded multivariate time-series data to a low-dimensional latent space 610. Predict the next time value of the coded multivariate time series data in space 615, the predicted next value and random noise are mapped back to the input space, the predicted distribution of the next time point of the multivariate time series data 620 provides a sample and outputs one or more of the time series predictions based on the predicted distribution sample 625. [Selection diagram] FIG. 6
机译:待解决的问题:提供用于多变量时间序列建模和预测的计算设备、计算机安装方法和计算机可读存储介质。时间序列建模和预测的计算机实现方法600对时间序列数据的多个输入605进行编码,将编码的多元时间序列数据映射到低维潜在空间,并将编码的多元时间序列数据映射到低维潜在空间610。预测空间615中编码的多元时间序列数据的下一时间值,预测的下一值和随机噪声映射回输入空间,多元时间序列数据620的下一时间点的预测分布提供样本,并基于预测分布样本625输出一个或多个时间序列预测。[选择图]图6

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号