Dissertation/ Thesis

Mixed order hyper-networks for function approximation and optimisation

Λεπτομέρειες βιβλιογραφικής εγγραφής
Τίτλος: Mixed order hyper-networks for function approximation and optimisation
Συγγραφείς: Swingler, Kevin
Συνεισφορές: Smith, Leslie, Hussain, Amir
Στοιχεία εκδότη: University of Stirling, 2016.
Έτος έκδοσης: 2016
Θεματικοί όροι: 006.3, Neural networks, Optimisation, Machine Learning, Estimation of Distribution Algorithms, Computers--Data processing, Computer algorithms, Combinatorial analysis--Data processing
Περιγραφή: Many systems take inputs, which can be measured and sometimes controlled, and outputs, which can also be measured and which depend on the inputs. Taking numerous measurements from such systems produces data, which may be used to either model the system with the goal of predicting the output associated with a given input (function approximation, or regression) or of finding the input settings required to produce a desired output (optimisation, or search). Approximating or optimising a function is central to the field of computational intelligence. There are many existing methods for performing regression and optimisation based on samples of data but they all have limitations. Multi layer perceptrons (MLPs) are universal approximators, but they suffer from the black box problem, which means their structure and the function they implement is opaque to the user. They also suffer from a propensity to become trapped in local minima or large plateaux in the error function during learning. A regression method with a structure that allows models to be compared, human knowledge to be extracted, optimisation searches to be guided and model complexity to be controlled is desirable. This thesis presents such as method. This thesis presents a single framework for both regression and optimisation: the mixed order hyper network (MOHN). A MOHN implements a function f:{-1,1}^n →R to arbitrary precision. The structure of a MOHN makes the ways in which input variables interact to determine the function output explicit, which allows human insights and complexity control that are very difficult in neural networks with hidden units. The explicit structure representation also allows efficient algorithms for searching for an input pattern that leads to a desired output. A number of learning rules for estimating the weights based on a sample of data are presented along with a heuristic method for choosing which connections to include in a model. Several methods for searching a MOHN for inputs that lead to a desired output are compared. Experiments compare a MOHN to an MLP on regression tasks. The MOHN is found to achieve a comparable level of accuracy to an MLP but suffers less from local minima in the error function and shows less variance across multiple training trials. It is also easier to interpret and combine from an ensemble. The trade-off between the fit of a model to its training data and that to an independent set of test data is shown to be easier to control in a MOHN than an MLP. A MOHN is also compared to a number of existing optimisation methods including those using estimation of distribution algorithms, genetic algorithms and simulated annealing. The MOHN is able to find optimal solutions in far fewer function evaluations than these methods on tasks selected from the literature.
Τύπος εγγράφου: Thesis Or Dissertation
Γλώσσα: English
Σύνδεσμος πρόσβασης: https://ethos.blethoshyku.com/concern/thesis_or_dissertations/714665
Αριθμός Καταχώρησης: edsble.714665
Βάση Δεδομένων: British Library EThOS
FullText Text:
  Availability: 0
CustomLinks:
  – Url: https://ethos.blethoshyku.com/concern/thesis_or_dissertations/714665
    Name: EDS - British Library EThOS (ns324271)
    Category: fullText
    Text: View record in EThOS
Header DbId: edsble
DbLabel: British Library EThOS
An: edsble.714665
RelevancyScore: 937
AccessLevel: 3
PubType: Dissertation/ Thesis
PubTypeId: dissertation
PreciseRelevancyScore: 937.178283691406
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Mixed order hyper-networks for function approximation and optimisation
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Swingler%2C+Kevin%22">Swingler, Kevin</searchLink>
– Name: Author
  Label: Contributors
  Group: Au
  Data: Smith, Leslie<br />Hussain, Amir
– Name: Publisher
  Label: Publisher Information
  Group: PubInfo
  Data: University of Stirling, 2016.
– Name: DatePubCY
  Label: Publication Year
  Group: Date
  Data: 2016
– Name: Subject
  Label: Subject Terms
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22006%2E3%22">006.3</searchLink><br /><searchLink fieldCode="DE" term="%22Neural+networks%22">Neural networks</searchLink><br /><searchLink fieldCode="DE" term="%22Optimisation%22">Optimisation</searchLink><br /><searchLink fieldCode="DE" term="%22Machine+Learning%22">Machine Learning</searchLink><br /><searchLink fieldCode="DE" term="%22Estimation+of+Distribution+Algorithms%22">Estimation of Distribution Algorithms</searchLink><br /><searchLink fieldCode="DE" term="%22Computers--Data+processing%22">Computers--Data processing</searchLink><br /><searchLink fieldCode="DE" term="%22Computer+algorithms%22">Computer algorithms</searchLink><br /><searchLink fieldCode="DE" term="%22Combinatorial+analysis--Data+processing%22">Combinatorial analysis--Data processing</searchLink>
– Name: Abstract
  Label: Description
  Group: Ab
  Data: Many systems take inputs, which can be measured and sometimes controlled, and outputs, which can also be measured and which depend on the inputs. Taking numerous measurements from such systems produces data, which may be used to either model the system with the goal of predicting the output associated with a given input (function approximation, or regression) or of finding the input settings required to produce a desired output (optimisation, or search). Approximating or optimising a function is central to the field of computational intelligence. There are many existing methods for performing regression and optimisation based on samples of data but they all have limitations. Multi layer perceptrons (MLPs) are universal approximators, but they suffer from the black box problem, which means their structure and the function they implement is opaque to the user. They also suffer from a propensity to become trapped in local minima or large plateaux in the error function during learning. A regression method with a structure that allows models to be compared, human knowledge to be extracted, optimisation searches to be guided and model complexity to be controlled is desirable. This thesis presents such as method. This thesis presents a single framework for both regression and optimisation: the mixed order hyper network (MOHN). A MOHN implements a function f:{-1,1}^n →R to arbitrary precision. The structure of a MOHN makes the ways in which input variables interact to determine the function output explicit, which allows human insights and complexity control that are very difficult in neural networks with hidden units. The explicit structure representation also allows efficient algorithms for searching for an input pattern that leads to a desired output. A number of learning rules for estimating the weights based on a sample of data are presented along with a heuristic method for choosing which connections to include in a model. Several methods for searching a MOHN for inputs that lead to a desired output are compared. Experiments compare a MOHN to an MLP on regression tasks. The MOHN is found to achieve a comparable level of accuracy to an MLP but suffers less from local minima in the error function and shows less variance across multiple training trials. It is also easier to interpret and combine from an ensemble. The trade-off between the fit of a model to its training data and that to an independent set of test data is shown to be easier to control in a MOHN than an MLP. A MOHN is also compared to a number of existing optimisation methods including those using estimation of distribution algorithms, genetic algorithms and simulated annealing. The MOHN is able to find optimal solutions in far fewer function evaluations than these methods on tasks selected from the literature.
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Thesis Or Dissertation
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: URL
  Label: Access URL
  Group: URL
  Data: <link linkTarget="URL" linkTerm="https://ethos.blethoshyku.com/concern/thesis_or_dissertations/714665" linkWindow="_blank">https://ethos.blethoshyku.com/concern/thesis_or_dissertations/714665</link>
– Name: AN
  Label: Accession Number
  Group: ID
  Data: edsble.714665
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=edsble&AN=edsble.714665
RecordInfo BibRecord:
  BibEntity:
    Languages:
      – Text: English
    Subjects:
      – SubjectFull: 006.3
        Type: general
      – SubjectFull: Neural networks
        Type: general
      – SubjectFull: Optimisation
        Type: general
      – SubjectFull: Machine Learning
        Type: general
      – SubjectFull: Estimation of Distribution Algorithms
        Type: general
      – SubjectFull: Computers--Data processing
        Type: general
      – SubjectFull: Computer algorithms
        Type: general
      – SubjectFull: Combinatorial analysis--Data processing
        Type: general
    Titles:
      – TitleFull: Mixed order hyper-networks for function approximation and optimisation
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Swingler, Kevin
      – PersonEntity:
          Name:
            NameFull: Smith, Leslie
      – PersonEntity:
          Name:
            NameFull: Hussain, Amir
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 01
              Type: published
              Y: 2016
          Identifiers:
            – Type: issn-locals
              Value: edsble
ResultId 1