Combining GMM, Jensen's inequality and BIC for speaker indexing

He, Qianhua; Yang, Junrong; Li, Y.-X.; He, Jun; Zhang, X.-Y.; Li, W.

doi:10.1049/el.2010.3003

Cited by 3 publications

(2 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…(2) 对于 π < θ r 2π 的情况, 方程 ∂P∆ ∂θr = 0 的根为: [19] . 首先, 设置任意满足 (15) 和 (17) 的初始值α r 和δ r , 并给出三个相位估计试探值θ 1…”

Section: 相位搜索unclassified

Research on the active cancellation technology of spectrum symbiotic interference

Guo¹,

Lin²,

Zhao³

et al. 2022

Sci. Sin.-Inf.

View full text Add to dashboard Cite

In this article, an active spectrum symbiotic interference (SSI) cancellation architecture is proposed for spectrum symbiotic wireless communication systems. On this basis, with the aim of minimizing the residual reception power, a joint SSI cancellation algorithm based on time-amplitude-phase three-dimensional search is proposed, which actively utilizes the characteristics of transmitted signal in the spectrum symbiotic wireless communication systems to adjust, reconstruct and cancel the SSI from the aspects of time, amplitude and phase. Theoretical and simulation analysis show that the proposed architecture and algorithm can suppress the SSI to the noise level, and have lower space complexity and time complexity.

show abstract

Section: 相位搜索unclassified

Research on the active cancellation technology of spectrum symbiotic interference

Guo¹,

Lin²,

Zhao³

et al. 2022

Sci. Sin.-Inf.

View full text Add to dashboard Cite

show abstract

“…BIC on the other hand provides similar classification performance while providing a lower number of needed mixtures. AIC and BIC are also commonly used when employing GMMs for a variety of other classification tasks [13]- [15].…”

Section: B Gaussian Mixture Models (Gmms)mentioning

confidence: 99%

Segmented Mixture-of-Gaussian Classification for Hyperspectral Image Analysis

Prasad

Cui

et al. 2014

IEEE Geosci. Remote Sensing Lett.

View full text Add to dashboard Cite

Abstract-The same high dimensionality of hyperspectral imagery that facilitates detection of subtle differences in spectral response due to differing chemical composition also hinders the deployment of traditional statistical pattern-classification procedures, particularly when relatively few training samples are available. Traditional approaches to addressing this issue, which typically employ dimensionality reduction based on either projection or feature selection, are at best suboptimal for hyperspectral classification tasks. A divide-and-conquer algorithm is proposed to exploit the high correlation between successive spectral bands and the resulting block-diagonal correlation structure to partition the hyperspectral space into approximately independent subspaces. Subsequently, dimensionality reduction based on a graph-theoretic localitypreserving discriminant analysis is combined with classification driven by Gaussian mixture models independently in each subspace. The locality-preserving discriminant analysis preserves the potentially multimodal statistical structure of the data, which the Gaussian mixture model classifier learns in the reduced-dimensional subspace. Experimental results demonstrate that the proposed system significantly outperforms traditional classification approaches, even when few training samples are employed.

show abstract