Common framework of certain reinforcement schedules

Author

Pacut, Andrzej

Author_Institution

Fac. of Electron. & Inf. Technol., Warsaw Univ. of Technol., Poland

Volume

3

fYear

1998

fDate

4-9 May 1998

Firstpage

2004

Abstract

We investigate reinforcement algorithms in a context of feedforward networks with gradient learning which use the smoothed output gradient estimators. The reduced network is introduced to avoid output redundancy. The adaptive critic element can be viewed as a network with smoothed output gradients, and the associative search elements the reduced network with smoothed output gradients. In this context, the adaptive critic element becomes a regular member of the family of adaptive critic designs

Keywords

adaptive control; adaptive systems; discrete time systems; feedforward neural nets; learning (artificial intelligence); neurocontrollers; adaptive critic element; associative search elements; feedforward networks; gradient learning; reduced network; reinforcement schedules; smoothed output gradient estimators; Adaptive systems; Control system synthesis; Control systems; Dynamic programming; Equations; Information resources; Information technology; Learning; Neural networks; Optimal control;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks Proceedings, 1998. IEEE World Congress on Computational Intelligence. The 1998 IEEE International Joint Conference on

Conference_Location

Anchorage, AK

ISSN

1098-7576

Print_ISBN

0-7803-4859-1

Type

conf

DOI

10.1109/IJCNN.1998.687167

Filename

687167