Semi-supervised learning in Spectral Dimensionality Reduction

Maryam Mehdizadeh

    Research output: ThesisMaster's Thesis

    205 Downloads (Pure)

    Abstract

    Biometric face data are essentially high dimensional data and as such are susceptible to the well-known problem of the curse of dimensionality when analyzed using machine learning techniques. Research has shown that biometric face data are non­linear in structure and, manifold learning methods are able to preserve the original non-linear structure of high dimensional data in lower dimensional space.
    Manifold learning methods suffer from two problems. First the generalization problem and second, the classification problem. In this dissertation, we will present Graph Embedding and Semi-supervised Graph Embedding as approaches to advance the applicability of manifold learning methods in dimensionality reduction of biometric face data for the purpose of recognition.
    Original languageEnglish
    QualificationMasters
    Awarding Institution
    • The University of Western Australia
    Supervisors/Advisors
    • Bennamoun, Mohammed, Supervisor
    • MacNish, Cara, Supervisor
    • Khan, Nazim, Supervisor
    Award date23 Sept 2016
    Publication statusUnpublished - 2016

    Fingerprint

    Dive into the research topics of 'Semi-supervised learning in Spectral Dimensionality Reduction'. Together they form a unique fingerprint.

    Cite this