English

Logarithmic Regret for Learning Linear Quadratic Regulators Efficiently

Machine Learning 2020-07-03 v2 Machine Learning

Abstract

We consider the problem of learning in Linear Quadratic Control systems whose transition parameters are initially unknown. Recent results in this setting have demonstrated efficient learning algorithms with regret growing with the square root of the number of decision steps. We present new efficient algorithms that achieve, perhaps surprisingly, regret that scales only (poly)logarithmically with the number of steps in two scenarios: when only the state transition matrix AA is unknown, and when only the state-action transition matrix BB is unknown and the optimal policy satisfies a certain non-degeneracy condition. On the other hand, we give a lower bound that shows that when the latter condition is violated, square root regret is unavoidable.

Keywords

Cite

@article{arxiv.2002.08095,
  title  = {Logarithmic Regret for Learning Linear Quadratic Regulators Efficiently},
  author = {Asaf Cassel and Alon Cohen and Tomer Koren},
  journal= {arXiv preprint arXiv:2002.08095},
  year   = {2020}
}

Comments

Accepted for presentation at International Conference on Machine Learning (ICML) 2020

R2 v1 2026-06-23T13:46:36.937Z