Full Gradient DQN Reinforcement Learning: A Provably Convergent Scheme
Machine Learning
2021-05-05 v3 Optimization and Control
Probability
Abstract
We analyze the DQN reinforcement learning algorithm as a stochastic approximation scheme using the o.d.e. (for 'ordinary differential equation') approach and point out certain theoretical issues. We then propose a modified scheme called Full Gradient DQN (FG-DQN, for short) that has a sound theoretical basis and compare it with the original scheme on sample problems. We observe a better performance for FG-DQN.
Cite
@article{arxiv.2103.05981,
title = {Full Gradient DQN Reinforcement Learning: A Provably Convergent Scheme},
author = {K. E. Avrachenkov and V. S. Borkar and H. P. Dolhare and K. Patil},
journal= {arXiv preprint arXiv:2103.05981},
year = {2021}
}