Preprocessing for Optimization of Probabilistic-Logic Models for Sequence Analysis

Research output: Contribution to journalConference articleResearchpeer-review

Abstract

A class of probabilistic-logic models is considered, which increases the expressibility from HMM's and SCFG's regular and context-free languages to, in principle, Turing complete languages. In general, such models are computationally far too complex for direct use, so optimization by pruning and approximation are needed. The first steps are taken towards a methodology for optimizing such models by approximations using auxiliary models for preprocessing or splitting them into submodels. Evaluation of such approximating models is challenging as authoritative test data may be sparse. On the other hand, the original complex models may be used for generating artificial evaluation data by efficient sampling, which can be used in the evaluation, although it does not constitute a foolproof test procedure. These models and evaluation processes are illustrated in the PRISM system developed by other authors, and we discuss their applicability and limitations.

Original languageEnglish
Book seriesLecture Notes in Computer Science
Pages (from-to)70-83
Number of pages14
ISSN0302-9743
Publication statusPublished - 2009
EventInternational Conference on Logic Programming 2009 - Pasadena, United States
Duration: 11 Jul 200917 Jul 2009
Conference number: 25

Conference

ConferenceInternational Conference on Logic Programming 2009
Number25
CountryUnited States
CityPasadena
Period11/07/200917/07/2009

Keywords

  • Logic programming
  • bio-informatics
  • probabilistic sequence analysis

Cite this

@inproceedings{f89df0b067b711de89b4000ea68e967b,
title = "Preprocessing for Optimization of Probabilistic-Logic Models for Sequence Analysis",
abstract = "A class of probabilistic-logic models is considered, which increases the expressibility from HMM's and SCFG's regular and context-free languages to, in principle, Turing complete languages. In general, such models are computationally far too complex for direct use, so optimization by pruning and approximation are needed. The first steps are taken towards a methodology for optimizing such models by approximations using auxiliary models for preprocessing or splitting them into submodels. Evaluation of such approximating models is challenging as authoritative test data may be sparse. On the other hand, the original complex models may be used for generating artificial evaluation data by efficient sampling, which can be used in the evaluation, although it does not constitute a foolproof test procedure. These models and evaluation processes are illustrated in the PRISM system developed by other authors, and we discuss their applicability and limitations.",
keywords = "Logic programming, bio-informatics, probabilistic sequence analysis",
author = "Henning Christiansen and Lassen, {Ole Torp}",
note = "Volumne: 5649",
year = "2009",
language = "English",
pages = "70--83",
journal = "Lecture Notes in Computer Science",
issn = "0302-9743",
publisher = "Physica-Verlag",

}

Preprocessing for Optimization of Probabilistic-Logic Models for Sequence Analysis. / Christiansen, Henning; Lassen, Ole Torp.

In: Lecture Notes in Computer Science, 2009, p. 70-83.

Research output: Contribution to journalConference articleResearchpeer-review

TY - GEN

T1 - Preprocessing for Optimization of Probabilistic-Logic Models for Sequence Analysis

AU - Christiansen, Henning

AU - Lassen, Ole Torp

N1 - Volumne: 5649

PY - 2009

Y1 - 2009

N2 - A class of probabilistic-logic models is considered, which increases the expressibility from HMM's and SCFG's regular and context-free languages to, in principle, Turing complete languages. In general, such models are computationally far too complex for direct use, so optimization by pruning and approximation are needed. The first steps are taken towards a methodology for optimizing such models by approximations using auxiliary models for preprocessing or splitting them into submodels. Evaluation of such approximating models is challenging as authoritative test data may be sparse. On the other hand, the original complex models may be used for generating artificial evaluation data by efficient sampling, which can be used in the evaluation, although it does not constitute a foolproof test procedure. These models and evaluation processes are illustrated in the PRISM system developed by other authors, and we discuss their applicability and limitations.

AB - A class of probabilistic-logic models is considered, which increases the expressibility from HMM's and SCFG's regular and context-free languages to, in principle, Turing complete languages. In general, such models are computationally far too complex for direct use, so optimization by pruning and approximation are needed. The first steps are taken towards a methodology for optimizing such models by approximations using auxiliary models for preprocessing or splitting them into submodels. Evaluation of such approximating models is challenging as authoritative test data may be sparse. On the other hand, the original complex models may be used for generating artificial evaluation data by efficient sampling, which can be used in the evaluation, although it does not constitute a foolproof test procedure. These models and evaluation processes are illustrated in the PRISM system developed by other authors, and we discuss their applicability and limitations.

KW - Logic programming

KW - bio-informatics

KW - probabilistic sequence analysis

M3 - Conference article

SP - 70

EP - 83

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

SN - 0302-9743

ER -