Learning Qualitative Models of Dynamic Systems

David T. Hau, Enrico W. Coiera

Research output: Contribution to journalArticleResearchpeer-review

Abstract

The automated construction of dynamic system models is an important application area for ILP. We describe a method that learns qualitative models from time-varying physiological signals. The goal is to understand the complexity of the learning task when faced with numerical data, what signal processing techniques are required, and how this affects learning. The qualitative representation is based on Kuipers' QSIM. The learning algorithm for model construction is based on Coiera's GENMODEL. We show that QSIM models are efficiently PAC learnable from positive examples only, and that GENMODEL is an ILP algorithm for efficiently constructing a QSIM model. We describe both GENMODEL which performs RLGG on qualitative states to learn a QSIM model, and the front-end processing and segmenting stages that transform a signal into a set of qualitative states. Next we describe results of experiments on data from six cardiac bypass patients. Useful models were obtained, representing both normal and abnormal physiological states. Model variation across time and across different levels of temporal abstraction and fault tolerance is explored. The assumption made by many previous workers that the abstraction of examples from data can be separated from the learning task is not supported by this study. Firstly, the effects of noise in the numerical data manifest themselves in the qualitative examples. Secondly, the models learned are directly dependent on the initial qualitative abstraction chosen.

LanguageEnglish
Pages177-211
Number of pages35
JournalMachine Learning
Volume26
Issue number2-3
Publication statusPublished - 1997
Externally publishedYes

Fingerprint

Dynamical systems
Inductive logic programming (ILP)
Fault tolerance
Learning algorithms
Signal processing
Processing
Experiments

Cite this

Hau, David T. ; Coiera, Enrico W. / Learning Qualitative Models of Dynamic Systems. In: Machine Learning. 1997 ; Vol. 26, No. 2-3. pp. 177-211.
@article{578ab2f5857a452eb9a38d5b910a5084,
title = "Learning Qualitative Models of Dynamic Systems",
abstract = "The automated construction of dynamic system models is an important application area for ILP. We describe a method that learns qualitative models from time-varying physiological signals. The goal is to understand the complexity of the learning task when faced with numerical data, what signal processing techniques are required, and how this affects learning. The qualitative representation is based on Kuipers' QSIM. The learning algorithm for model construction is based on Coiera's GENMODEL. We show that QSIM models are efficiently PAC learnable from positive examples only, and that GENMODEL is an ILP algorithm for efficiently constructing a QSIM model. We describe both GENMODEL which performs RLGG on qualitative states to learn a QSIM model, and the front-end processing and segmenting stages that transform a signal into a set of qualitative states. Next we describe results of experiments on data from six cardiac bypass patients. Useful models were obtained, representing both normal and abnormal physiological states. Model variation across time and across different levels of temporal abstraction and fault tolerance is explored. The assumption made by many previous workers that the abstraction of examples from data can be separated from the learning task is not supported by this study. Firstly, the effects of noise in the numerical data manifest themselves in the qualitative examples. Secondly, the models learned are directly dependent on the initial qualitative abstraction chosen.",
author = "Hau, {David T.} and Coiera, {Enrico W.}",
year = "1997",
language = "English",
volume = "26",
pages = "177--211",
journal = "Machine Learning",
issn = "0885-6125",
publisher = "Springer, Springer Nature",
number = "2-3",

}

Hau, DT & Coiera, EW 1997, 'Learning Qualitative Models of Dynamic Systems', Machine Learning, vol. 26, no. 2-3, pp. 177-211.

Learning Qualitative Models of Dynamic Systems. / Hau, David T.; Coiera, Enrico W.

In: Machine Learning, Vol. 26, No. 2-3, 1997, p. 177-211.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Learning Qualitative Models of Dynamic Systems

AU - Hau, David T.

AU - Coiera, Enrico W.

PY - 1997

Y1 - 1997

N2 - The automated construction of dynamic system models is an important application area for ILP. We describe a method that learns qualitative models from time-varying physiological signals. The goal is to understand the complexity of the learning task when faced with numerical data, what signal processing techniques are required, and how this affects learning. The qualitative representation is based on Kuipers' QSIM. The learning algorithm for model construction is based on Coiera's GENMODEL. We show that QSIM models are efficiently PAC learnable from positive examples only, and that GENMODEL is an ILP algorithm for efficiently constructing a QSIM model. We describe both GENMODEL which performs RLGG on qualitative states to learn a QSIM model, and the front-end processing and segmenting stages that transform a signal into a set of qualitative states. Next we describe results of experiments on data from six cardiac bypass patients. Useful models were obtained, representing both normal and abnormal physiological states. Model variation across time and across different levels of temporal abstraction and fault tolerance is explored. The assumption made by many previous workers that the abstraction of examples from data can be separated from the learning task is not supported by this study. Firstly, the effects of noise in the numerical data manifest themselves in the qualitative examples. Secondly, the models learned are directly dependent on the initial qualitative abstraction chosen.

AB - The automated construction of dynamic system models is an important application area for ILP. We describe a method that learns qualitative models from time-varying physiological signals. The goal is to understand the complexity of the learning task when faced with numerical data, what signal processing techniques are required, and how this affects learning. The qualitative representation is based on Kuipers' QSIM. The learning algorithm for model construction is based on Coiera's GENMODEL. We show that QSIM models are efficiently PAC learnable from positive examples only, and that GENMODEL is an ILP algorithm for efficiently constructing a QSIM model. We describe both GENMODEL which performs RLGG on qualitative states to learn a QSIM model, and the front-end processing and segmenting stages that transform a signal into a set of qualitative states. Next we describe results of experiments on data from six cardiac bypass patients. Useful models were obtained, representing both normal and abnormal physiological states. Model variation across time and across different levels of temporal abstraction and fault tolerance is explored. The assumption made by many previous workers that the abstraction of examples from data can be separated from the learning task is not supported by this study. Firstly, the effects of noise in the numerical data manifest themselves in the qualitative examples. Secondly, the models learned are directly dependent on the initial qualitative abstraction chosen.

UR - http://www.scopus.com/inward/record.url?scp=0031069041&partnerID=8YFLogxK

M3 - Article

VL - 26

SP - 177

EP - 211

JO - Machine Learning

T2 - Machine Learning

JF - Machine Learning

SN - 0885-6125

IS - 2-3

ER -