Presentation 2001/3/16
Effectiveness of Gauss-Sigmoid Neural Network in Reinforcement Learning with Continuous Inputs
Shin'ichi MAEHARA, Masanori SUGISAKA, Katsunari SHIBATA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Boyan et al. has pointed out that the combination of reinforcement learning and Sigmoid-based neural network sometimes leads to instability of the learning. In this paper, it is proposed that a Gauss-Sigmoid neural network, in which continuous input signals are put into a Sigmoid-based neural network through a RBF network, is utilized for reinforcement learning. It is confirmed through simulations of the hill-car task that the learning results are far better in the case of the Gauss-Sigmoid neural network, than in the case of the Sigmoid-based neural network. Also it is comfirmed that the Gauss-Sigmoid neural network obtains a global representation in the hidden layer through learning and the generalization can work effectively on the representation.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Reinforcement learning / Gauss-Sigmoid Neural Nerwork / RBF / Hill-car problem / generalization
Paper # NC2000-166
Date of Issue

Conference Information
Committee NC
Conference Date 2001/3/16(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Effectiveness of Gauss-Sigmoid Neural Network in Reinforcement Learning with Continuous Inputs
Sub Title (in English)
Keyword(1) Reinforcement learning
Keyword(2) Gauss-Sigmoid Neural Nerwork
Keyword(3) RBF
Keyword(4) Hill-car problem
Keyword(5) generalization
1st Author's Name Shin'ichi MAEHARA
1st Author's Affiliation Depertment of Electrical and Electronic Engineering, Oita University()
2nd Author's Name Masanori SUGISAKA
2nd Author's Affiliation Depertment of Electrical and Electronic Engineering, Oita University
3rd Author's Name Katsunari SHIBATA
3rd Author's Affiliation Depertment of Electrical and Electronic Engineering, Oita University
Date 2001/3/16
Paper # NC2000-166
Volume (vol) vol.100
Number (no) 688
Page pp.pp.-
#Pages 8
Date of Issue