Natural systems, from climate and ocean circulation to organisms and cells, involve complex dynamics extending over multiple spatio-temporal scales. Centuries old efforts to forecast these dynamics have been hindered by the high dimensionality and chaotic behavior of these systems. Forecasting of high-dimensional systems is also a challenge for data science applications ranging from biology to finance.

We introduce a data-driven forecasting method for high dimensional, chaotic
systems using Long-Short Term Memory (LSTM) recurrent neural networks (RNNs). The proposed LSTM-RNNs perform inference of high dimensional dynamical systems in their reduced order space and are shown to be an effective set of non-linear approximators of their attractor. We demonstrate the forecasting performance of the LSTM and compare it with Gaussian processes (GPs) in time series obtained from the Lorenz 96 system, the Kuramoto-Sivashinsky equation and a prototype climate model. The LSTM networks outperform the GPs in short-term forecasting accuracy in all applications considered. A hybrid architecture, extending the LSTM with a mean stochastic model (MSM-LSTM), is proposed to ensure statistics of the attractor are captured in the long term. This novel hybrid method is fully data-driven and extends the forecasting capabilities of LSTM networks for chaotic dynamical systems.

Time: 14:00 - 14:30

Track 5
organised by SSS

Speaker: Pantelis Vlachas