Tailored Forecasting from Short Time Series via Meta-learning
This addresses forecasting challenges for systems with limited historical data, such as in scientific or industrial applications, though it is incremental as it builds on existing meta-learning and reservoir computing methods.
The paper tackles the problem of forecasting dynamical systems from short time-series data by introducing METAFORS, a meta-learning approach that generalizes knowledge from related systems with longer histories. It demonstrates reliable prediction of short-term dynamics and long-term statistics on simulated chaotic systems, even when test and related systems differ substantially.
Machine learning models can effectively forecast dynamical systems from time-series data, but they typically require large amounts of past data, making forecasting particularly challenging for systems with limited history. To overcome this, we introduce Meta-learning for Tailored Forecasting using Related Time Series (METAFORS), which generalizes knowledge across systems to enable forecasting in data-limited scenarios. By learning from a library of models trained on longer time series from potentially related systems, METAFORS builds and initializes a model tailored to short time-series data from the system of interest. Using a reservoir computing implementation and testing on simulated chaotic systems, we demonstrate that METAFORS can reliably predict both short-term dynamics and long-term statistics without requiring contextual labels. We see this even when test and related systems exhibit substantially different behaviors, highlighting METAFORS' strengths in data-limited scenarios.