Title :
Realizing Undelayed N-step TD prediction with neural networks
Author_Institution :
Fac. of Comput., Univ. of Latvia, Riga, Latvia
Abstract :
There exist various techniques to extend reinforcement learning algorithms, e.g., eligibility traces and planning. In this paper, an approach is proposed, which combines several extension techniques, such as using eligibility-like traces, using approximators as value functions and exploiting the model of the environment. The obtained method, `Undelayed n-step TD prediction´ (TD-P), has produced competitive results when put in conditions of not fully observable environment.
Keywords :
learning (artificial intelligence); neural nets; eligibility planning; eligibility traces; neural networks; realizing undelayed N-step TD prediction; reinforcement learning algorithms; Computer networks; Delay; Dynamic programming; Machine learning; Multi-layer neural network; Multilayer perceptrons; Neural networks; Predictive models;
Conference_Titel :
MELECON 2010 - 2010 15th IEEE Mediterranean Electrotechnical Conference
Conference_Location :
Valletta
Print_ISBN :
978-1-4244-5793-9
DOI :
10.1109/MELCON.2010.5476332
