Stochastic dynamics learning with state-space systems
This provides theoretical foundations for reliable generative modeling of temporal data in reservoir computing, addressing both deterministic and stochastic regimes.
This work advances reservoir computing theory by establishing that fading memory and solution stability hold generically in state-space systems, even without strict contractivity conditions, and proposes a novel distributional perspective for stochastic echo states based on attractor dynamics on probability distributions.
This work advances the theoretical foundations of reservoir computing (RC) by providing a unified treatment of fading memory and the echo state property (ESP) in both deterministic and stochastic settings. We investigate state-space systems, a central model class in time series learning, and establish that fading memory and solution stability hold generically -- even in the absence of the ESP -- offering a robust explanation for the empirical success of RC models without strict contractivity conditions. In the stochastic case, we critically assess stochastic echo states, proposing a novel distributional perspective rooted in attractor dynamics on the space of probability distributions, which leads to a rich and coherent theory. Our results extend and generalize previous work on non-autonomous dynamical systems, offering new insights into causality, stability, and memory in RC models. This lays the groundwork for reliable generative modeling of temporal data in both deterministic and stochastic regimes.