Delay-SDE-net: A deep learning approach for time series modelling with memory and uncertainty estimates

Abstract

To model time series accurately is important within a wide range of fields. As the world is generally too complex to be modelled exactly, it is often meaningful to assess the probability of a dynamical system to be in a specific state. This paper presents the Delay-SDE-net, a neural network model based on stochastic delay differential equations (SDDEs). The use of SDDEs with multiple delays as modelling framework makes it a suitable model for time series with memory effects, as it includes memory through previous states of the system. The stochastic part of the Delay-SDE-net provides a basis for estimating uncertainty in modelling, and is split into two neural networks to account for aleatoric and epistemic uncertainty. The uncertainty is provided instantly, making the model suitable for applications where time is sparse. We derive the theoretical error of the Delay-SDE-net and analyze the convergence rate numerically. At comparisons with similar models, the Delay-SDE-net has consistently the best performance, both in predicting time series values and uncertainties.