• ETRI Journal
• /
• 제36권1호
• /
• pp.167-170
• /
• 2014

In this letter, we propose an unsupervised framework for speech noise reduction based on the recent development of low-rank and sparse matrix decomposition. The proposed framework directly separates the speech signal from noisy speech by decomposing the noisy speech spectrogram into three submatrices: the noise structure matrix, the clean speech structure matrix, and the residual noise matrix. Evaluations on the Noisex-92 dataset show that the proposed method achieves a signal-to-distortion ratio approximately 2.48 dB and 3.23 dB higher than that of the robust principal component analysis method and the non-negative matrix factorization method, respectively, when the input SNR is -5 dB.

• ETRI Journal
• /
• 제39권1호
• /
• pp.21-29
• /
• 2017

This paper addresses the problem of unsupervised speech separation based on robust non-negative matrix factorization (RNMF) with ${\beta}$-divergence, when neither speech nor noise training data is available beforehand. We propose a robust version of non-negative matrix factorization, inspired by the recently developed sparse and low-rank decomposition, in which the data matrix is decomposed into the sum of a low-rank matrix and a sparse matrix. Efficient multiplicative update rules to minimize the ${\beta}$-divergence-based cost function are derived. A convolutional extension of the proposed algorithm is also proposed, which considers the time dependency of the non-negative noise bases. Experimental speech separation results show that the proposed convolutional RNMF successfully separates the repeating time-varying spectral structures from the magnitude spectrum of the mixture, and does so without any prior training.

• 대한수학회논문집
• /
• 제28권1호
• /
• pp.1-9
• /
• 2013

We consider the sets of matrix ordered pairs which satisfy extremal properties with respect to Boolean rank inequalities of matrices over nonbinary Boolean algebra. We characterize linear operators that preserve these sets of matrix ordered pairs as the form of $T(X)=PXP^T$ with some permutation matrix P.

• 대한수학회지
• /
• 제56권2호
• /
• pp.507-521
• /
• 2019

For any $m{\times}n$ nonbinary Boolean matrix A, its spanning column rank is the minimum number of the columns of A that spans its column space. We have a characterization of spanning column rank-1 nonbinary Boolean matrices. We investigate the linear operators that preserve the spanning column ranks of matrices over the nonbinary Boolean algebra. That is, for a linear operator T on $m{\times}n$ nonbinary Boolean matrices, it preserves all spanning column ranks if and only if there exist an invertible nonbinary Boolean matrix P of order m and a permutation matrix Q of order n such that T(A) = PAQ for all $m{\times}n$ nonbinary Boolean matrix A. We also obtain other characterizations of the (spanning) column rank preserver.

• 제어로봇시스템학회논문지
• /
• 제15권1호
• /
• pp.67-71
• /
• 2009

This paper presents a rank-constrained linear matrix inequality (LMI) approach to the design of a multi-objective controller such as $H_2/H_{\infty}$ control. Multi-objective control is formulated as an LMI optimization problem with a nonconvex rank condition, which is imposed on the controller gain matirx not Lyapunov matrices. With this rank-constrained formulation, we can expect to reduce conservatism because we can use separate Lyapunov matrices for different control objectives. An iterative penalty method is applied to solve this rank-constrained LMI optimization problem. Numerical experiments are performed to illustrate the proposed method.

• Journal of the Korean Data and Information Science Society
• /
• 제7권1호
• /
• pp.79-85
• /
• 1996

This paper will show that the rank of the model matrix of a closed, n compartmental model with k sinks is n-k. This statement will be extended to include open compartmental models as a part of theorem.

• 대한수학회지
• /
• 제51권1호
• /
• pp.113-123
• /
• 2014

The term rank of a matrix A over a semiring $\mathcal{S}$ is the least number of lines (rows or columns) needed to include all the nonzero entries in A. In this paper, we characterize linear operators that preserve the sets of matrix ordered pairs which satisfy extremal properties with respect to term rank inequalities of matrices over nonbinary Boolean semirings.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1997년도 한국자동제어학술회의논문집; 한국전력공사 서울연수원; 17-18 Oct. 1997
• /
• pp.105-108
• /
• 1997

This paper proposes a new approach of balanced model reduction using matrix pencil. The algorithm presented in this paper is to convert full-rank high-order system into rank-deficient system using perturbation made by matrix pencil method. Then the system can be truncated to a low-order system that we want via balanced realization. We discuss the comparison with other methods and the various observations by simulations.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권12호
• /
• pp.6043-6062
• /
• 2019

Low-rank matrix decomposition has shown its capability in many applications such as image in-painting, de-noising, background reconstruction and defect detection etc. In this paper, we consider the texture background of rail track images and the sparse foreground of the defects to construct a low-rank matrix decomposition model with block sparsity for defect inspection of rail tracks, which jointly minimizes the nuclear norm and the 2-1 norm. Similar to ADM, an alternative method is proposed in this study to solve the optimization problem. After image decomposition, the defect areas in the resulting low-rank image will form dark stripes that horizontally cross the entire image, indicating the preciselocations of the defects. Finally, a two-stage defect extraction method is proposed to locate the defect areas. The experimental results of the two datasets show that our algorithm achieved better performance compared with other methods.

• 한국정보통신학회논문지
• /
• 제25권10호
• /
• pp.1317-1322
• /
• 2021

영상 신호에는 비지역적 유사성이 존재하므로 임의의 조각 영상 즉, 패치(patch)에 대해 유사한 패치들을 모아 구성한 패치행렬은 낮은 계수값(rank)을 갖는 특성이 있다. 백색 잡음이 섞인 영상으로 구성된 패치행렬은 원 영상에 비해 높은 계수값을 갖게 된다. 이 행렬에 대해 저 계수의 근사 행렬을 구하면 영상 속의 잡음을 제거할 수 있다. 본 논문에서는 기준 패치의 유사 패치들을 이용한 패치행렬 구성 방법과 패치행렬에 대한 저 계수 행렬 근사 방법 및 이를 이용한 영상 복원 방법으로 구성된 영상 잡음 제거 방식을 제안한다. 또한 모의실험을 통해 제안된 방식의 잡음 제거 성능을 최신 4가지 방법들과 비교하여 그 우수성을 보인다.