Learning Beyond Experience: Generalizing to Unseen State Space with Reservoir Computing
This addresses the challenge of generalization in data-driven modeling of dynamical systems for researchers and practitioners, though it appears incremental as it builds on existing reservoir computing methods.
The paper tackled the problem of machine learning models struggling to generalize to unseen regions of state space without structural priors, and demonstrated that reservoir computing with a multiple-trajectory training scheme can achieve out-of-domain generalization, capturing behavior in unobserved basins of multistable dynamical systems.
Machine learning techniques offer an effective approach to modeling dynamical systems solely from observed data. However, without explicit structural priors -- built-in assumptions about the underlying dynamics -- these techniques typically struggle to generalize to aspects of the dynamics that are poorly represented in the training data. Here, we demonstrate that reservoir computing -- a simple, efficient, and versatile machine learning framework often used for data-driven modeling of dynamical systems -- can generalize to unexplored regions of state space without explicit structural priors. First, we describe a multiple-trajectory training scheme for reservoir computers that supports training across a collection of disjoint time series, enabling effective use of available training data. Then, applying this training scheme to multistable dynamical systems, we show that RCs trained on trajectories from a single basin of attraction can achieve out-of-domain generalization by capturing system behavior in entirely unobserved basins.