Discriminative State Space Models
Abstract
In this paper, we introduce and analyze Discriminative State Space Models for
forecasting non-stationary time series. We provide data-dependent generalization
guarantees for learning these models based on recently introduced notion of discrep-
ancy. We provide an in-depth analysis of complexity of such models. Finally, we
also study generalization ability of several structural risk minimization approaches
to this problem and provide efficient implementation for one of them which is
based on a convex objective.