English

Hierarchical model-based policy optimization: from actions to action sequences and back

Machine Learning 2020-01-03 v2 Artificial Intelligence Optimization and Control Machine Learning

Abstract

We develop a normative framework for hierarchical model-based policy optimization based on applying second-order methods in the space of all possible state-action paths. The resulting natural path gradient performs policy updates in a manner which is sensitive to the long-range correlational structure of the induced stationary state-action densities. We demonstrate that the natural path gradient can be computed exactly given an environment dynamics model and depends on expressions akin to higher-order successor representations. In simulation, we show that the priorization of local policy updates in the resulting policy flow indeed reflects the intuitive state-space hierarchy in several toy problems.

Keywords

Cite

@article{arxiv.1912.01448,
  title  = {Hierarchical model-based policy optimization: from actions to action sequences and back},
  author = {Daniel McNamee},
  journal= {arXiv preprint arXiv:1912.01448},
  year   = {2020}
}

Comments

NeurIPS 2019 Optimization Foundations of Reinforcement Learning Workshop. v2: typos fixed, minor edits for improved clarity

R2 v1 2026-06-23T12:34:28.831Z