Learning from many trajectories
Abstract
We initiate a study of supervised learning from many independent sequences ("trajectories") of non-independent covariates, reflecting tasks in sequence modeling, control, and reinforcement learning. Conceptually, our multi-trajectory setup sits between two traditional settings in statistical learning theory: learning from independent examples and learning from a single auto-correlated sequence. Our conditions for efficient learning generalize the former setting--trajectories must be non-degenerate in ways that extend standard requirements for independent examples. Notably, we do not require that trajectories be ergodic, long, nor strictly stable. For linear least-squares regression, given -dimensional examples produced by trajectories, each of length , we observe a notable change in statistical efficiency as the number of trajectories increases from a few (namely ) to many (namely ). Specifically, we establish that the worst-case error rate of this problem is whenever . Meanwhile, when , we establish a (sharp) lower bound of on the worst-case error rate, realized by a simple, marginally unstable linear dynamical system. A key upshot is that, in domains where trajectories regularly reset, the error rate eventually behaves as if all of the examples were independent, drawn from their marginals. As a corollary of our analysis, we also improve guarantees for the linear system identification problem.
Cite
@article{arxiv.2203.17193,
title = {Learning from many trajectories},
author = {Stephen Tu and Roy Frostig and Mahdi Soltanolkotabi},
journal= {arXiv preprint arXiv:2203.17193},
year = {2023}
}