Knowledge-based Quadratic Discriminant Analysis for phonetic classification

Huang, Heyun; Liu, Yang; Bosch, L.F.M. ten; Cranen, B.; Boves, L.W.J.

doi:10.1109/icassp.2012.6288831

Cited by 4 publications

(2 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The use of second-order polynomial features improves the performance of phone classification [1], and automatic speech recognition (ASR) systems [2,3]. Higher-order polynomial features [4,5], as a natural extension to the second-order polynomial features, might convey discriminative information which adds to the first-/second-order polynomial features and might improve ASR performance.…”

Section: Introductionmentioning

confidence: 99%

Training log-linear acoustic models in higher-order polynomial feature space for speech recognition

et al. 2013

Self Cite

View full text Add to dashboard Cite

The use of higher-order polynomial acoustic features can improve the performance of automatic speech recognition. However, the dimensionality of the polynomial representation can be prohibitively large, making the training of acoustic models using polynomial features for large vocabulary ASR systems infeasible. This paper presents an iterative polynomial training framework for acoustic modeling, which recursively expands the current acoustic features into their second-order polynomial feature space. In each recursion the dimensionality is reduced by a linear projection, such that increasingly higher order polynomial information is incorporated while keeping the dimensionality of the acoustic models constant. Experimental results obtained for a large-vocabulary continuous speech recognition task show that the proposed method outperforms conventional mixture models.

show abstract

Section: Introductionmentioning

confidence: 99%

Training log-linear acoustic models in higher-order polynomial feature space for speech recognition

et al. 2013

Self Cite

View full text Add to dashboard Cite

show abstract

“…Therefore, their preliminary analyses suggest that RF and methods similar in nature to RF may be more useful than other methods to classify samples based on MS data. Compared to Linear Discriminant Analysis (LDA) and Quadratic Discriminant Analysis (QDA) methods (Huang et al, 2012), RF has the advantage of not requiring the number of variables used to be less than the number of subjects in the study, which is a clear advantage for the analysis of MS data as the number of m/z versus intensity data points is very large. In addition, RF is able to handle interactions among variables.…”

Section: Literature Reviewmentioning

confidence: 99%

Spectral Domain Features for Ovarian Cancer Data Analysis

Seddik

Hassan

Fakhreldein³

2013

Journal of Computer Science

View full text Add to dashboard Cite

The early detection of cancer is crucial for successful treatment. Medical researchers have investigated a number of early-diagnosis techniques. Recently, they have discovered that some cancers affect the concentration of certain molecules in the blood, which allows early diagnosis by analyzing the blood mass spectrum. Researchers have developed several techniques for the analysis of the mass-spectrum curve analysis and used them for the detection of prostate, ovarian, breast, bladder, pancreatic, kidney, liver and colon cancers. In this study we propose a new technique that uses the spectral domain features such as wavelet transform and Fourier transform for the analysis of the ovarian cancer data to differentiate between normal and patients with malignant cancer. We used two different classifiers for the original data, the first one is a feed forward artificial neural network classifier which gave a sensitivity of 96%, specificity of 88% and accuracy of 94%. The second used classifier is the linear discriminant analysis classifier which separated the cancer from healthy samples with sensitivity of 79%, specificity of 75% and accuracy of about 81%. After transforming the data to the spectral domain using the Fourier transform the performance was degraded. The experimental results showed that the performance of the wavelet transform based system was superior to other techniques as it gave a sensitivity of 98%, specificity of 96% and accuracy of 95%.

show abstract