PaperBank

Download Here
Second Author HomePage

Abstract:
Convergence for iterative reinforcement learning algorithms like TD(0) depends on the sampling strategy for the transitions. However, in practical applications it is convenient to take transition data from arbitrary sources without losing convergence. In this paper we investigate the problem of repeated synchronous updates based on a xed set of transitions. Our main theorem yields sufficient conditions of convergence for combinations of reinforcement learning algorithms and linear...

Keywords: Reinforcement Learning, TD(0), Convergence

@misc{ schoknecht03convergent,
author = "R. Schoknecht and A. Merke",
title = "Convergent combinations of reinforcement learning with function approximation",
text = "R. Schoknecht and A. Merke. Convergent combinations of reinforcement learning
with function approximation. In Advances in Neural Information Processing
Systems, volume 15, 2003.",
year = "2003",
url = "citeseer.ist.psu.edu/schoknecht03convergent.html" }

	Reinforcement Learning and Artificial Intelligence (RLAI)
	R. Schoknecht and A. Merke. Convergent combinations of reinforcement learning with function approximation. In Advances in Neural Information Processing Systems, volume 15, 2003.