Perception control with improved expectation learning through multilayered neural networks

Sherwin Guirnaldo, Keigo Watanabe, Kiyotaka Izumi, Kazuo Kiguchi

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

In this paper, we investigate the viability of multilayered neural network (NN)-based extension of a conventional "perception" control concept. The perception process selects and completes the information from the system to be controlled before passing it to the controlling agent so that control is not lost when sensory information from the system is incomplete. The perception process produces an expectation of the next set of information to be received from the system. The expectation is used to replace missing parts of the information received and it also influences the next perception. In the existing work, each of the expectation elements is linearly acquired such that the expectation tells only the dominant information in the recent past, i.e., this approach has no capability to sense the trend and the dynamics in the information. This handicap could become a serious problem when the perception process is applied to real physical systems. Here, we introduce an extension of the perception control process by using a radial basis function (RBF) feedforward NN to learn the trend and the dynamics in the information and produce the expectation of the next observation. Through some simulation comparisons, we show that the proposed RBFNN-based method is better than the existing one.

Original languageEnglish
Pages (from-to)1582-1587
Number of pages6
JournalIEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
Volume34
Issue number3
DOIs
Publication statusPublished - Jun 2004
Externally publishedYes

Fingerprint

Neural networks
Feedforward neural networks

Keywords

  • Expectation
  • Perception control
  • Q-learning
  • RBF neural networks
  • Sensor interruption

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Artificial Intelligence
  • Human-Computer Interaction

Cite this

Perception control with improved expectation learning through multilayered neural networks. / Guirnaldo, Sherwin; Watanabe, Keigo; Izumi, Kiyotaka; Kiguchi, Kazuo.

In: IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics, Vol. 34, No. 3, 06.2004, p. 1582-1587.

Research output: Contribution to journalArticle

@article{aa250c1902614b4a88df29ba9935e454,
title = "Perception control with improved expectation learning through multilayered neural networks",
abstract = "In this paper, we investigate the viability of multilayered neural network (NN)-based extension of a conventional {"}perception{"} control concept. The perception process selects and completes the information from the system to be controlled before passing it to the controlling agent so that control is not lost when sensory information from the system is incomplete. The perception process produces an expectation of the next set of information to be received from the system. The expectation is used to replace missing parts of the information received and it also influences the next perception. In the existing work, each of the expectation elements is linearly acquired such that the expectation tells only the dominant information in the recent past, i.e., this approach has no capability to sense the trend and the dynamics in the information. This handicap could become a serious problem when the perception process is applied to real physical systems. Here, we introduce an extension of the perception control process by using a radial basis function (RBF) feedforward NN to learn the trend and the dynamics in the information and produce the expectation of the next observation. Through some simulation comparisons, we show that the proposed RBFNN-based method is better than the existing one.",
keywords = "Expectation, Perception control, Q-learning, RBF neural networks, Sensor interruption",
author = "Sherwin Guirnaldo and Keigo Watanabe and Kiyotaka Izumi and Kazuo Kiguchi",
year = "2004",
month = "6",
doi = "10.1109/TSMCB.2004.825922",
language = "English",
volume = "34",
pages = "1582--1587",
journal = "IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics",
issn = "1083-4419",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "3",

}

TY - JOUR

T1 - Perception control with improved expectation learning through multilayered neural networks

AU - Guirnaldo, Sherwin

AU - Watanabe, Keigo

AU - Izumi, Kiyotaka

AU - Kiguchi, Kazuo

PY - 2004/6

Y1 - 2004/6

N2 - In this paper, we investigate the viability of multilayered neural network (NN)-based extension of a conventional "perception" control concept. The perception process selects and completes the information from the system to be controlled before passing it to the controlling agent so that control is not lost when sensory information from the system is incomplete. The perception process produces an expectation of the next set of information to be received from the system. The expectation is used to replace missing parts of the information received and it also influences the next perception. In the existing work, each of the expectation elements is linearly acquired such that the expectation tells only the dominant information in the recent past, i.e., this approach has no capability to sense the trend and the dynamics in the information. This handicap could become a serious problem when the perception process is applied to real physical systems. Here, we introduce an extension of the perception control process by using a radial basis function (RBF) feedforward NN to learn the trend and the dynamics in the information and produce the expectation of the next observation. Through some simulation comparisons, we show that the proposed RBFNN-based method is better than the existing one.

AB - In this paper, we investigate the viability of multilayered neural network (NN)-based extension of a conventional "perception" control concept. The perception process selects and completes the information from the system to be controlled before passing it to the controlling agent so that control is not lost when sensory information from the system is incomplete. The perception process produces an expectation of the next set of information to be received from the system. The expectation is used to replace missing parts of the information received and it also influences the next perception. In the existing work, each of the expectation elements is linearly acquired such that the expectation tells only the dominant information in the recent past, i.e., this approach has no capability to sense the trend and the dynamics in the information. This handicap could become a serious problem when the perception process is applied to real physical systems. Here, we introduce an extension of the perception control process by using a radial basis function (RBF) feedforward NN to learn the trend and the dynamics in the information and produce the expectation of the next observation. Through some simulation comparisons, we show that the proposed RBFNN-based method is better than the existing one.

KW - Expectation

KW - Perception control

KW - Q-learning

KW - RBF neural networks

KW - Sensor interruption

UR - http://www.scopus.com/inward/record.url?scp=2942549040&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=2942549040&partnerID=8YFLogxK

U2 - 10.1109/TSMCB.2004.825922

DO - 10.1109/TSMCB.2004.825922

M3 - Article

VL - 34

SP - 1582

EP - 1587

JO - IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics

JF - IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics

SN - 1083-4419

IS - 3

ER -