Learning with imperfect perception

W. Wen, M. Yokoo

Research output: Contribution to conferencePaper

Abstract

Machine learning algorithms which adopt a state space representation usually assume perfect knowledge of what state the system is currently in. This is to guarantee that rewards and penalties are correctly assigned to the responsible state. This assumption, however, does not hold in most real world learning problems due to imperfect perception. In this paper estimation and control theory is used to classify the systems depending on the observability of the system states. This observability determines whether the optimal control strategy of a particular system can be learned, A novel approach based on enhancing the observability is used to deal with perceptual aliasing problem. In order to learn to perceive, the perception actions are directly integrated into the control actions. An example is shown and further applications to robot learning is discussed.

Original languageEnglish
Pages209-218
Number of pages10
Publication statusPublished - Dec 1 1994
EventProceedings of the 4th IEEE Workshop on Neural Networks for Signal Processing (NNSP'94) - Ermioni, GREECE
Duration: Sep 6 1994Sep 8 1994

Other

OtherProceedings of the 4th IEEE Workshop on Neural Networks for Signal Processing (NNSP'94)
CityErmioni, GREECE
Period9/6/949/8/94

Fingerprint

Observability
Robot learning
Control theory
Learning algorithms
Learning systems

All Science Journal Classification (ASJC) codes

  • Signal Processing
  • Software
  • Electrical and Electronic Engineering

Cite this

Wen, W., & Yokoo, M. (1994). Learning with imperfect perception. 209-218. Paper presented at Proceedings of the 4th IEEE Workshop on Neural Networks for Signal Processing (NNSP'94), Ermioni, GREECE, .

Learning with imperfect perception. / Wen, W.; Yokoo, M.

1994. 209-218 Paper presented at Proceedings of the 4th IEEE Workshop on Neural Networks for Signal Processing (NNSP'94), Ermioni, GREECE, .

Research output: Contribution to conferencePaper

Wen, W & Yokoo, M 1994, 'Learning with imperfect perception' Paper presented at Proceedings of the 4th IEEE Workshop on Neural Networks for Signal Processing (NNSP'94), Ermioni, GREECE, 9/6/94 - 9/8/94, pp. 209-218.
Wen W, Yokoo M. Learning with imperfect perception. 1994. Paper presented at Proceedings of the 4th IEEE Workshop on Neural Networks for Signal Processing (NNSP'94), Ermioni, GREECE, .
Wen, W. ; Yokoo, M. / Learning with imperfect perception. Paper presented at Proceedings of the 4th IEEE Workshop on Neural Networks for Signal Processing (NNSP'94), Ermioni, GREECE, .10 p.
@conference{143611cba534436aa79c080a0ec133bc,
title = "Learning with imperfect perception",
abstract = "Machine learning algorithms which adopt a state space representation usually assume perfect knowledge of what state the system is currently in. This is to guarantee that rewards and penalties are correctly assigned to the responsible state. This assumption, however, does not hold in most real world learning problems due to imperfect perception. In this paper estimation and control theory is used to classify the systems depending on the observability of the system states. This observability determines whether the optimal control strategy of a particular system can be learned, A novel approach based on enhancing the observability is used to deal with perceptual aliasing problem. In order to learn to perceive, the perception actions are directly integrated into the control actions. An example is shown and further applications to robot learning is discussed.",
author = "W. Wen and M. Yokoo",
year = "1994",
month = "12",
day = "1",
language = "English",
pages = "209--218",
note = "Proceedings of the 4th IEEE Workshop on Neural Networks for Signal Processing (NNSP'94) ; Conference date: 06-09-1994 Through 08-09-1994",

}

TY - CONF

T1 - Learning with imperfect perception

AU - Wen, W.

AU - Yokoo, M.

PY - 1994/12/1

Y1 - 1994/12/1

N2 - Machine learning algorithms which adopt a state space representation usually assume perfect knowledge of what state the system is currently in. This is to guarantee that rewards and penalties are correctly assigned to the responsible state. This assumption, however, does not hold in most real world learning problems due to imperfect perception. In this paper estimation and control theory is used to classify the systems depending on the observability of the system states. This observability determines whether the optimal control strategy of a particular system can be learned, A novel approach based on enhancing the observability is used to deal with perceptual aliasing problem. In order to learn to perceive, the perception actions are directly integrated into the control actions. An example is shown and further applications to robot learning is discussed.

AB - Machine learning algorithms which adopt a state space representation usually assume perfect knowledge of what state the system is currently in. This is to guarantee that rewards and penalties are correctly assigned to the responsible state. This assumption, however, does not hold in most real world learning problems due to imperfect perception. In this paper estimation and control theory is used to classify the systems depending on the observability of the system states. This observability determines whether the optimal control strategy of a particular system can be learned, A novel approach based on enhancing the observability is used to deal with perceptual aliasing problem. In order to learn to perceive, the perception actions are directly integrated into the control actions. An example is shown and further applications to robot learning is discussed.

UR - http://www.scopus.com/inward/record.url?scp=0028727455&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0028727455&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:0028727455

SP - 209

EP - 218

ER -