On Phonemes As Cognitive Components of Speech

Ling Feng, Lars Kai Hansen

    Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

    142 Downloads (Pure)

    Abstract

    COgnitive Component Analysis (COCA) defined as the process of unsupervised grouping of data such that the ensuing group structure is well-aligned with that resulting from human cognitive activity, has been explored on phoneme data. Statistical regularities have been revealed at multiple time scales. The basic features are 25-dimensional short time (20ms) melfrequency weighted cepstral coefficients. Features are integrated by means of stacking to obtain features at longer time scales. Energy based sparsification is carried out to achieve sparse representations. Our hypothesis is ecological: we assume that features that essentially independent in a context defined ensemble can be efficiently coded using a sparse independent component representation. This means that supervised and unsupervised learning should result in similar representations. We indeed find that supervised and unsupervised learning seem to identify similar representations, here, measured by the classification similarity.
    Original languageEnglish
    Title of host publicationThe 1st IAPR Workshop on Cognitive Information Processing
    Publication date2008
    Publication statusPublished - 2008
    EventThe 1st IAPR Workshop on Cognitive Information Processing - Santorini, Greece
    Duration: 1 Jan 2008 → …

    Conference

    ConferenceThe 1st IAPR Workshop on Cognitive Information Processing
    CitySantorini, Greece
    Period01/01/2008 → …

    Keywords

    • Unsupervised Learning
    • Cognitive Component Analysis
    • Phoneme Classification
    • Supervised Learning

    Cite this

    Feng, L., & Hansen, L. K. (2008). On Phonemes As Cognitive Components of Speech. In The 1st IAPR Workshop on Cognitive Information Processing
    Feng, Ling ; Hansen, Lars Kai. / On Phonemes As Cognitive Components of Speech. The 1st IAPR Workshop on Cognitive Information Processing. 2008.
    @inproceedings{1086b1ecb24444dbb3b461fd867f055d,
    title = "On Phonemes As Cognitive Components of Speech",
    abstract = "COgnitive Component Analysis (COCA) defined as the process of unsupervised grouping of data such that the ensuing group structure is well-aligned with that resulting from human cognitive activity, has been explored on phoneme data. Statistical regularities have been revealed at multiple time scales. The basic features are 25-dimensional short time (20ms) melfrequency weighted cepstral coefficients. Features are integrated by means of stacking to obtain features at longer time scales. Energy based sparsification is carried out to achieve sparse representations. Our hypothesis is ecological: we assume that features that essentially independent in a context defined ensemble can be efficiently coded using a sparse independent component representation. This means that supervised and unsupervised learning should result in similar representations. We indeed find that supervised and unsupervised learning seem to identify similar representations, here, measured by the classification similarity.",
    keywords = "Unsupervised Learning, Cognitive Component Analysis, Phoneme Classification, Supervised Learning",
    author = "Ling Feng and Hansen, {Lars Kai}",
    year = "2008",
    language = "English",
    booktitle = "The 1st IAPR Workshop on Cognitive Information Processing",

    }

    Feng, L & Hansen, LK 2008, On Phonemes As Cognitive Components of Speech. in The 1st IAPR Workshop on Cognitive Information Processing. The 1st IAPR Workshop on Cognitive Information Processing, Santorini, Greece, 01/01/2008.

    On Phonemes As Cognitive Components of Speech. / Feng, Ling; Hansen, Lars Kai.

    The 1st IAPR Workshop on Cognitive Information Processing. 2008.

    Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

    TY - GEN

    T1 - On Phonemes As Cognitive Components of Speech

    AU - Feng, Ling

    AU - Hansen, Lars Kai

    PY - 2008

    Y1 - 2008

    N2 - COgnitive Component Analysis (COCA) defined as the process of unsupervised grouping of data such that the ensuing group structure is well-aligned with that resulting from human cognitive activity, has been explored on phoneme data. Statistical regularities have been revealed at multiple time scales. The basic features are 25-dimensional short time (20ms) melfrequency weighted cepstral coefficients. Features are integrated by means of stacking to obtain features at longer time scales. Energy based sparsification is carried out to achieve sparse representations. Our hypothesis is ecological: we assume that features that essentially independent in a context defined ensemble can be efficiently coded using a sparse independent component representation. This means that supervised and unsupervised learning should result in similar representations. We indeed find that supervised and unsupervised learning seem to identify similar representations, here, measured by the classification similarity.

    AB - COgnitive Component Analysis (COCA) defined as the process of unsupervised grouping of data such that the ensuing group structure is well-aligned with that resulting from human cognitive activity, has been explored on phoneme data. Statistical regularities have been revealed at multiple time scales. The basic features are 25-dimensional short time (20ms) melfrequency weighted cepstral coefficients. Features are integrated by means of stacking to obtain features at longer time scales. Energy based sparsification is carried out to achieve sparse representations. Our hypothesis is ecological: we assume that features that essentially independent in a context defined ensemble can be efficiently coded using a sparse independent component representation. This means that supervised and unsupervised learning should result in similar representations. We indeed find that supervised and unsupervised learning seem to identify similar representations, here, measured by the classification similarity.

    KW - Unsupervised Learning

    KW - Cognitive Component Analysis

    KW - Phoneme Classification

    KW - Supervised Learning

    M3 - Article in proceedings

    BT - The 1st IAPR Workshop on Cognitive Information Processing

    ER -

    Feng L, Hansen LK. On Phonemes As Cognitive Components of Speech. In The 1st IAPR Workshop on Cognitive Information Processing. 2008