Systems and methods for time series forecasting

Abstract

Systems and methods for providing a neural network system for time series forecasting are described. A time series dataset that includes datapoints at a plurality of timestamps in an observed space is received. A first state-space model of a dynamical system underlying the time series dataset is provided. The first state-space model includes a non-parametric latent transition model. One or more latent variables of a latent space for the time series dataset are determined using the neural network system based on the first state-space model. A first prediction result for the time series dataset is provided by the neural network system based on the estimated latent variables.

What is claimed is: 
     
         1 . A method of providing a neural network system for time series forecasting, comprising:
 receiving, via an input interface, a time series dataset that includes datapoints at a plurality of timestamps in an observed space;   providing a first state-space model of a dynamical system underlying the time series dataset,
 wherein the first state-space model includes a first non-parametric transition function for generating one or more latent variables of a latent space for the time series dataset; 
   determining, using the neural network system based on the first state-space model, one or more estimated latent variables; and   providing, using the neural network system, a first prediction result for the time series dataset based on the estimated latent variables.   
     
     
         2 . The method of  claim 1 , wherein each of the latent variables is identifiable from a corresponding observed data. 
     
     
         3 . The method of  claim 1 , wherein the first nonparametric transition function has a first input including the one or more parent time-lagged variables of the latent factor and a second input including a first noise. 
     
     
         4 . The method of  claim 3 , wherein the first nonparametric transition function has a third input including one or more time-varying change factors. 
     
     
         5 . The method of  claim 4 , wherein the one or more time-varying change factors are associated with a second nonparametric transition function. 
     
     
         6 . The method of  claim 5 , wherein the second nonparametric transition function includes an input including a second noise, and
 wherein the first noise and the second noise are mutually independent.   
     
     
         7 . The method of  claim 1 , wherein the neural network system includes:
 an encoder configured to determine the one or more estimated latent variables;   an auxiliary predictor configured to generate a first latent-space prediction result based on the one or more estimated latent variables; and   a decoder configured to transform the first latent-space prediction result to the first prediction result in the observed space.   
     
     
         8 . A non-transitory machine-readable medium comprising a plurality of machine-readable instructions which, when executed by one or more processors, are adapted to cause the one or more processors to perform a method comprising:
 receiving, via an input interface, a time series dataset that includes datapoints at a plurality of timestamps in an observed space;   providing a first state-space model of a dynamical system underlying the time series dataset,
 wherein the first state-space model includes a first non-parametric transition function for generating one or more latent variables of a latent space for the time series dataset; 
   determining, using a neural network system based on the first state-space model, one or more estimated latent variables; and   providing, using the neural network system, a first prediction result for the time series dataset based on the estimated latent variables.   
     
     
         9 . The non-transitory machine-readable medium of  claim 8 , wherein each of the latent variables is identifiable from a corresponding observed data. 
     
     
         10 . The non-transitory machine-readable medium of  claim 8 , wherein the first nonparametric transition function has a first input including the one or more parent time-lagged variables of the latent factor and a second input including a first noise. 
     
     
         11 . The non-transitory machine-readable medium of  claim 10 , wherein the first nonparametric transition function has a third input including one or more time-varying change factors. 
     
     
         12 . The non-transitory machine-readable medium of  claim 11 , wherein the one or more time-varying change factors are associated with a second nonparametric transition function. 
     
     
         13 . The non-transitory machine-readable medium of  claim 12 , wherein the second nonparametric transition function has an input including a second noise, and
 wherein the first noise and the second noise are mutually independent.   
     
     
         14 . The non-transitory machine-readable medium of  claim 8 , wherein the neural network system includes:
 an encoder configured to determine the one or more estimated latent variables;   an auxiliary predictor configured to generate a first latent-space prediction result based on the one or more estimated latent variables; and   a decoder configured to transform the first latent-space prediction result to the first prediction result in the observed space.   
     
     
         15 . A system, comprising:
 a non-transitory memory; and   one or more hardware processors coupled to the non-transitory memory and configured to read instructions from the non-transitory memory to cause the system to perform a method comprising:
 receiving, via an input interface, a time series dataset that includes datapoints at a plurality of timestamps in an observed space; 
 providing a first state-space model of a dynamical system underlying the time series dataset,
 wherein the first state-space model includes a first non-parametric transition function for generating one or more latent variables of a latent space for the time series dataset; 
 
 determining, using a neural network system based on the first state-space model, one or more estimated latent variables; and 
 providing, using the neural network system, a first prediction result for the time series dataset based on the estimated latent variables. 
   
     
     
         16 . The system of  claim 15 , wherein each of the latent variables is identifiable from a corresponding observed data. 
     
     
         17 . The system of  claim 15 , wherein the first nonparametric transition function has a first input including the one or more parent time-lagged variables of the latent factor and a second input including a first noise. 
     
     
         18 . The system of  claim 17 , wherein the first nonparametric transition function has a third input including one or more time-varying change factors. 
     
     
         19 . The system of  claim 18 , wherein the one or more time-varying change factors are associated with a second nonparametric transition function. 
     
     
         20 . The system of  claim 19 , wherein the second nonparametric transition function has an input including a second noise, and
 wherein the first noise and the second noise are mutually independent.