Umbrella Reinforcement Learning -- computationally efficient tool for hard non-linear problems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Nuzhin, Egor E., Brilliantov, Nikolai V.
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912249294094336
author Nuzhin, Egor E.
Brilliantov, Nikolai V.
author_facet Nuzhin, Egor E.
Brilliantov, Nikolai V.
contents We report a novel, computationally efficient approach for solving hard nonlinear problems of reinforcement learning (RL). Here we combine umbrella sampling, from computational physics/chemistry, with optimal control methods. The approach is realized on the basis of neural networks, with the use of policy gradient. It outperforms, by computational efficiency and implementation universality, all available state-of-the-art algorithms, in application to hard RL problems with sparse reward, state traps and lack of terminal states. The proposed approach uses an ensemble of simultaneously acting agents, with a modified reward which includes the ensemble entropy, yielding an optimal exploration-exploitation balance.
format Preprint
id arxiv_https___arxiv_org_abs_2411_14117
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Umbrella Reinforcement Learning -- computationally efficient tool for hard non-linear problems
Nuzhin, Egor E.
Brilliantov, Nikolai V.
Machine Learning
Artificial Intelligence
I.2.6; I.2.8
We report a novel, computationally efficient approach for solving hard nonlinear problems of reinforcement learning (RL). Here we combine umbrella sampling, from computational physics/chemistry, with optimal control methods. The approach is realized on the basis of neural networks, with the use of policy gradient. It outperforms, by computational efficiency and implementation universality, all available state-of-the-art algorithms, in application to hard RL problems with sparse reward, state traps and lack of terminal states. The proposed approach uses an ensemble of simultaneously acting agents, with a modified reward which includes the ensemble entropy, yielding an optimal exploration-exploitation balance.
title Umbrella Reinforcement Learning -- computationally efficient tool for hard non-linear problems
topic Machine Learning
Artificial Intelligence
I.2.6; I.2.8
url https://arxiv.org/abs/2411.14117