ICML 2018oral55 citations

Scalable Bilinear Pi Learning Using State and Action Features

Yichen Chen, Lihong Li, Mengdi Wang

Abstract

Approximate linear programming (ALP) represents one of the major algorithmic families to solve large-scale Markov decision processes (MDP). In this work, we study a primal-dual formulation of the ALP, and develop a scalable, model-free algorithm called bilinear $\pi$ learning for reinforcement learning when a sampling oracle is provided. This algorithm enjoys a number of advantages. First, it adopts linear and bilinear models to represent the high-dimensional value function and state-action distributions, respectively, using given state and action features. Its run-time complexity depends on the number of features, not the size of the underlying MDPs. Second, it operates in a fully online fashion without having to store any sample, thus having minimal memory footprint. Third, we prove that it is sample-efficient, solving for the optimal policy to high precision with a sample complexity linear in the dimension of the parameter space.

BibTeX
@InProceedings{pmlr-v80-chen18e,
  title = 	 {Scalable Bilinear Pi Learning Using State and Action Features},
  author =       {Chen, Yichen and Li, Lihong and Wang, Mengdi},
  booktitle = 	 {Proceedings of the 35th International Conference on Machine Learning},
  pages = 	 {834--843},
  year = 	 {2018},
  editor = 	 {Dy, Jennifer and Krause, Andreas},
  volume = 	 {80},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {10--15 Jul},
  publisher =    {PMLR},
  pdf = 	 {http://proceedings.mlr.press/v80/chen18e/chen18e.pdf},
  url = 	 {https://proceedings.mlr.press/v80/chen18e.html},
  abstract = 	 {Approximate linear programming (ALP) represents one of the major algorithmic families to solve large-scale Markov decision processes (MDP). In this work, we study a primal-dual formulation of the ALP, and develop a scalable, model-free algorithm called bilinear $\pi$ learning for reinforcement learning when a sampling oracle is provided. This algorithm enjoys a number of advantages. First, it adopts linear and bilinear models to represent the high-dimensional value function and state-action distributions, respectively, using given state and action features. Its run-time complexity depends on the number of features, not the size of the underlying MDPs. Second, it operates in a fully online fashion without having to store any sample, thus having minimal memory footprint. Third, we prove that it is sample-efficient, solving for the optimal policy to high precision with a sample complexity linear in the dimension of the parameter space.}
}