Application of novelty filter to segmentation of speech

Haltsonen, S.; Jalanko, Mahir; Bry, K.-J.; Kohonen, Teuvo

doi:10.1109/icassp.1978.1170488

Cited by 4 publications

(3 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These unitized representations may both influence, and be influenced by, hippocampal patterns characterized by a yet stronger form of invariance, expressed in allocentric and conceptual representations (Buzsáki & Moser, 2013;O'Keefe & Dostrovsky, 1971;Pennatz, 2015;Quiroga, 2012). We stress that these two computational hypotheses are not mutually exclusive and that PER may harbor additional computational functions, such as novelty filtering (Haltsonen, Jalanko, Bry, & Kohonen, 1978). Returning to the "gating" function of the PER, both computational paradigms allow to hypothesize that unitized representations, stored as such in the PER and its closely connected structures, may be evoked and reinstated when sufficient bottom-up sensory evidence is presented through lowerlevel cortical areas to allow completion of the unitized pattern, which then acts as a prerequisite for further transmission into the hippocampal system.…”

Section: Discussionmentioning

confidence: 87%

Reconciling the object and spatial processing views of the perirhinal cortex through task‐relevant unitization

et al. 2021

View full text Add to dashboard Cite

The perirhinal cortex is situated on the border between sensory association cortex and the hippocampal formation. It serves an important function as a transition area between the sensory neocortex and the medial temporal lobe. While the perirhinal cortex has traditionally been associated with object coding and the "what" pathway of the temporal lobe, current evidence suggests a broader function of the perirhinal cortex in solving feature ambiguity and processing complex stimuli. Besides fulfilling functions in object coding, recent neurophysiological findings in freely moving rodents indicate that the perirhinal cortex also contributes to spatial and contextual processing beyond individual sensory modalities. Here, we address how these two opposing views on perirhinal cortex-the object-centered and spatial-contextual processing hypotheses-may be reconciled. The perirhinal cortex is consistently recruited when different features can be merged perceptually or conceptually into a single entity. Features that are unitized in these entities include object information from multiple sensory domains, reward associations, semantic features and spatial/contextual associations. We propose that the same perirhinal network circuits can be flexibly deployed for multiple cognitive functions, such that the perirhinal cortex performs similar unitization operations on different types of information, depending on behavioral demands and ranging from the object-related domain to spatial, contextual and semantic information.

show abstract

Section: Discussionmentioning

confidence: 87%

Reconciling the object and spatial processing views of the perirhinal cortex through task‐relevant unitization

et al. 2021

View full text Add to dashboard Cite

show abstract

“…The pattern vectors were formed by concatenation of two 15-element spectra taken at a temporal interval of 30 ms. Such a concatenation was needed in an integrated system described elsewhere (2)(3)(4) Separate sdbspaces were defined for each of the nasals In n p/ but due to their poor mutual discrimination their recognition accuracies will be reported here as one group. The result, however, is better than if they were included in the same subspace.…”

Section: Tee Practical Applicationmentioning

confidence: 99%

“…Its applicability to the recognition of phonemes from continuous speech has been demonstrated in a few previous works (2)(3)(4). In this paper, the subspace method is developed further to yield results of practical value.…”

Section: Introductionmentioning

confidence: 98%

Spectral classification of phonemes by learning subspaces

Kohonen¹,

Németh²,

Bry³

et al.

ICASSP '79. IEEE International Conference on Acoustics, Speech, and Signal Processing

Self Cite

View full text Add to dashboard Cite

A new pattern recognition scheme with learning ability is introduced, and its application to the labeling of phonemes is reported. The basic classification algorithm is known as the subspace method in which classes of patterns are defined as linear vector subspaces spanned by the prototypes, and the class affiliation of an unknown pattern vector is decided by comparison of its orthogonal projections on the various subspaces. This method is here modified in two ways. In one of them, the prototype patterns are selected conditionally according to classification results obtained during training. In the second modification the subspaces are rotated in proper directions in the training procedure, depending on the classification results. By means of these methods, for the average accuracy of classification with 15 phonemic classes from continuous Finnish speech, a value of about 80 per cent was obtained.

show abstract

Application of the subspace method to speech recognition

Jalanko¹,

Kohonen²

ICASSP '78. IEEE International Conference on Acoustics, Speech, and Signal Processing

View full text Add to dashboard Cite

In this paper, the subspace method of pattern recognition, in which method classification is decided by the largest projection of an unknown pattern vector onto subspaces corresponding to different classes, is applied to the recognition of continuous Finnish speech.Classification is based on phonemic power spectra produced by an analog filter bank. When compared, e.g., with the nearest-neighbor method and the method of direction cosines, the advantages of the subspace method are an improved stability of classification and a more balanced total classification accuracy of the different phonemic classes with respect to their relative frequencies of occurrence. The efficient spanning of the subspaces as well as their mutual orthogonalization are discussed. Furthermore, the close relationship between the phonemic labeling and segmentation when using the subspace method is pointed Out.

show abstract

Application of novelty filter to segmentation of speech

Cited by 4 publications

References 3 publications

Reconciling the object and spatial processing views of the perirhinal cortex through task‐relevant unitization

Reconciling the object and spatial processing views of the perirhinal cortex through task‐relevant unitization

Spectral classification of phonemes by learning subspaces

Application of the subspace method to speech recognition

Contact Info

Product

Resources

About