State space probabilistic multi-time sequence prediction method based on graph neural network

Abstract

The invention discloses a state space probabilistic multi-time sequence prediction method based on a graph neural network, and the method comprises the steps: (1) obtaining a multi-time sequence, carrying out the preprocessing of the time sequence to construct a training set, and constructing a graph structure; (2) constructing a generation model for generating prior distribution and a time sequence of the hidden state according to a graph neural network and a multilayer perceptron, and constructing an inference network for generating approximate posteriori distribution of the hidden state according to the graph neural network and a recurrent neural network; (3) constructing a loss function according to the prior distribution and the approximate posteriori distribution of the implicit state, and optimizing a generation model and deducing parameters of the network by taking maximization of the loss function as a target; and (4) during application, obtaining the hidden state estimation of the to-be-predicted sequence at the latest moment by utilizing the inference network, then calculating to obtain the prior distribution of the hidden state by utilizing the generation model according to the hidden state estimation at the latest moment, and then calculating to obtain a time sequence observation estimation value according to the prior distribution of the hidden state.

Description

technical field [0001] The invention relates to the field of time series prediction, in particular to a state space probabilistic multi-time series prediction method based on graph neural network. Background technique [0002] Time series data widely exist in fields such as business, finance, smart cities, medical care, and environmental science. Time series forecasting technology can play an important role in applications such as data analysis, intelligent decision-making, and anomaly detection. In many scenarios, multiple homogeneous and related time series are observed and recorded simultaneously, such as sales of related commodities, prices of related stocks, traffic of urban roads, EEG, air quality in adjacent areas, etc. [0003] For such multiple time series, a simple forecasting method is to use the classic statistical model represented by ARIMA to independently model and forecast each time series. Although this method has the advantages of simple parameter solving ...

AI Technical Summary

Problems solved by technology

The advantage of this method is that it can use the strong and effective fitting ability of the neural network to model the nonlinear correlation between multiple time series, but the disadvantage is that it models the input-output mapping in a black box, which is easy to learn and helps to improve the prediction Spurious statistical correlation of accuracy, leading to poor generalization performance and difficulty in providing meaningful uncertainty estimates for prediction results

Images