English

Scale Invariant Monte Carlo under Linear Function Approximation with Curvature based step-size

Machine Learning 2022-05-31 v2

Abstract

We study the feature-scaled version of the Monte Carlo algorithm with linear function approximation. This algorithm converges to a scale-invariant solution, which is not unduly affected by states having feature vectors with large norms. The usual versions of the MCMC algorithm, obtained by minimizing the least-squares criterion, do not produce solutions that give equal importance to all states irrespective of feature-vector norm -- a requirement that may be critical in many reinforcement learning contexts. To speed up convergence in our algorithm, we introduce an adaptive step-size based on the curvature of the iterate convergence path -- a novelty that may be useful in more general optimization contexts as well. A key contribution of this paper is to prove convergence, in the presence of adaptive curvature based step-size and heavy-ball momentum. We provide rigorous theoretical guarantees and use simulations to demonstrate the efficacy of our ideas.

Keywords

Cite

@article{arxiv.2104.07361,
  title  = {Scale Invariant Monte Carlo under Linear Function Approximation with Curvature based step-size},
  author = {Rahul Madhavan and Hemanta Makwana},
  journal= {arXiv preprint arXiv:2104.07361},
  year   = {2022}
}

Comments

42 pages, 9 figures (9 pages main body with 5 figures)

R2 v1 2026-06-24T01:11:40.487Z