• 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.

• 청정기술
• /
• 제20권3호
• /
• pp.321-329
• /
• 2014

본 연구에서는 고수분 저등급 석탄을 저수분 석탄으로 만들기 위하여 실험실 규모의 회분식 스팀 유동층 건조기를 사용하여 수분이 약 26%함유된 인도네시아산 저등급 석탄을 5% 이하로 건조하였다. 일반적으로 이산화탄소 포집 및 저장기술(carbon capture and storage, CCS)은 $CO_2$를 재생하는 공정에서 $100{\sim}150^{\circ}C$의 스팀과 $CO_2$혼합가스를 배출한다. 이때 배출되는 가스의 열을 사용하여 저등급 석탄을 건조하는 것이 본 연구의 최종 목적이다. 이를 위하여 본 연구에서는 건조의 열원으로 스팀을 사용하고, 유동화 가스는 $CO_2$를 사용하여 저등급 석탄을 건조하였다. 연구에 사용한 스팀의 유량은 0.3~1.1 kg/hr, 온도는 $100~130^{\circ}C$, 석탄의 층높이는 9~25 cm로 변화시켰다. 건조 후 석탄의 특성 변화는 공업분석, 발열량분석 그리고 입자크기 분석을 통하여 확인하였다. 변수 실험을 수행한 결과 원탄의 건조속도는 스팀의 유량과 온도가 증가함에 따라 증가하였고, 층높이가 감소할수록 건조속도가 증가하였다.

• 정보와 통신
• /
• 제31권11호
• /
• pp.35-45
• /
• 2014

주성분 분석 기법(PCA)는 가장 널리 사용되는 데이터 차원 감소 (dimensionality reduction) 기법으로 알려져 있다. 하지만 데이터에 이상점 (outlier)가 존재하는 환경에서는 성능이 크게 저하된다는 단점을 가지고 있다. Rank-Sparsity(Robust PCA) 기법은 주어진 행렬을 low-rank 행렬과 저밀도(sparse)행렬의 합으로 분해하는 방식으로, 이상점이 많은 환경에서 PCA기법을 효과적으로 대체할 수 있는 알고리즘으로 알려져 있다. 본 고에서는 RPCA 기법을 간략히 소개하고, 그의 적용분야, 및 알고리즘에 관한 연구들을 대해서 알아본다.

• 한국컴퓨터정보학회논문지
• /
• 제27권5호
• /
• pp.77-84
• /
• 2022

본 논문에서는 새로운 비지도 특징 선별 기법을 제안한다. 기존 비지도 방식의 특징 선별 기법들은 특징을 선별하기 위해 가상의 레이블 데이터를 정하고 주어진 데이터를 이 레이블 데이터에 사영하는 회귀 분석 방식으로 특징을 선별하였다. 하지만 가상의 레이블은 데이터로부터 생성되기 때문에 사영된 공간이 비슷하게 형성될 수 있다. 따라서 기존의 방법들에서는 제한된 공간에서만 특징이 선택될 수 있었다. 이를 해소하기 위해 본 논문에서는 직교 사영과 저랭크 근사를 이용하여 특징을 선별한다. 이 문제를 해소하기 위해 가상의 레이블을 직교 사영하고 이 공간에 데이터를 사영할 수 있도록 한다. 이를 통해 더 주요한 특징 선별을 기대할 수 있다. 그리고 사영을 위한 변환 행렬에 저랭크 제한을 두어 더 효과적으로 저차원 공간의 특징을 선별할 수 있도록 한다. 이 목표를 달성하기 위해 본 논문에서는 비용 함수를 설계하고 효율적인 최적화 방법을 제안한다. 여섯 개의 데이터에 대한 실험 결과는 제안된 방법이 대부분의 경우 기존의 비지도 특징 선별 기법보다 좋은 성능을 보여주었다.

• Communications for Statistical Applications and Methods
• /
• 제5권3호
• /
• pp.707-722
• /
• 1998

This paper evaluates a new clinical trial designs for low infection rate disease. This type of sparse disease reaction makes the traditional two sample t-test or Wilcoxon rank-sum test inefficient compared to a new test suggested. The new test, which is based solely on the larger changes, is shown to be more effective than existing method by simulation for small samples. However, this test can be shown to be connected to the locally most powerful rank test under certain practical conditions. This design is motivated in testing the treatment effects in periodontal disease research.

• Environmental Engineering Research
• /
• 제25권4호
• /
• pp.522-528
• /
• 2020

To achieve the clean and efficient utilization of low-rank coal, the combustion and pollutant emission characteristics of chars from low-temperature and fast pyrolysis in a horizontal tube furnace were investigated in a TG-MS analyzer. According to the results, the combustion characteristic of chars was poorer than its parent coals. The temperature range of gaseous product release had a good agreement with that of TGA weight loss. Gaseous products of samples with high content of volatile were released earlier. The NO and NO₂ emissions of chars were lower than their parent coals. Coals of high rank (anthracite and sub-bituminous) released more NO and NO₂ than low rank coals of lignite, so were chars from coals of different ranks. SO₂ emissions of char samples were lower than parent coals and did not show obvious relationship with coal ranks.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2004년도 학술대회 논문집 정보 및 제어부문
• /
• pp.729-731
• /
• 2004

This paper deals with the design of a low order $H_{\infty}$ controller by using an iterative linear matrix inequality (LMI) method. The low order $H_{\infty}$ controller is represented in terms of LMIs with a rank condition. To solve the non-convex rank-constrained LMI problem, a linear penalty function is incorporated into the objective function so that minimizing the penalized objective function subject to LMIs amounts to a convex optimization problem. With an increasing sequence of the penalty parameter, the solution of the penalized optimization problem moves towards the feasible region of the original non-convex problem. The proposed algorithm is, therefore, convergent. Numerical experiments show the effectiveness of the proposed algorithm.

• 대한수학회지
• /
• 제60권6호
• /
• pp.1137-1169
• /
• 2023

Let C be a curve and V → C an orthogonal vector bundle of rank r. For r ≤ 6, the structure of V can be described using tensor, symmetric and exterior products of bundles of lower rank, essentially due to the existence of exceptional isomorphisms between Spin(r, ℂ) and other groups for these r. We analyze these structures in detail, and in particular use them to describe moduli spaces of orthogonal bundles. Furthermore, the locus of isotropic vectors in V defines a quadric subfibration Q_V ⊂ ℙV . Using familiar results on quadrics of low dimension, we exhibit isomorphisms between isotropic Quot schemes of V and certain ordinary Quot schemes of line subbundles. In particular, for r ≤ 6 this gives a method for enumerating the isotropic subbundles of maximal degree of a general V , when there are finitely many.

• 자원환경지질
• /
• 제32권1호
• /
• pp.83-91
• /
• 1999

Coal properties are controlled by the following two factors : One is the maceral components and the other is the degree of coalification. In other words, even if coals in question indicate the same degree of coalification, their chemical and physical properties considerably vary one another when their maceral components are different. It is well known that virtrinite reflectance is the best single criterion for the degree of coalification covering the whole range of coal rank. Some authors have recently insisted that sporinite fluorescence is more leliable coal rank parameter than vitrinite reflectance in case of low rank coals. In this paper, to examine the relince of sporinite fluorescence as coal rank parameter, fluidity analysis of coals is newly performed and the data are analyzed in comparision with those of virinite reflectance, sporinite fluorescence and maceral components. The results of this study are as follows; 1) Vitrinite reflectance becomes low when degradinite content is high within one columnar samples, and vice versa. 2) variation of vitrinite reflectance depend on degradinite content and on difference of roiginal plant. 3) In dealing with the Japanese paleogene coals, sporinite fluorescence is more reliable parameter indicating the degree of coalification than vitrinite reflectance. 4) Maximum fluidity increases exponetially in proportion to the increases of degradinite content.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권3호
• /
• pp.813-829
• /
• 2022

In multi-view subspace clustering, how to integrate the complementary information between perspectives to construct a unified representation is a critical problem. In the existing works, the unified representation is usually constructed in the original data space. However, when the data representation in each view is very diverse, the unified representation derived directly in the original data domain may lead to a huge information loss. To address this issue, different to the existing works, inspired by the latest revelation that the data across all perspectives have a very similar or close spectral block structure, we try to construct the unified representation in the spectral embedding domain. In this way, the complementary information across all perspectives can be fused into a unified representation with little information loss, since the spectral block structure from all views shares high consistency. In addition, to capture the global structure of data on each view with high accuracy and robustness both, we propose a novel low-rank approximation via the tight lower bound on the rank function. Finally, experimental results prove that, the proposed method has the effectiveness and robustness at the same time, compared with the state-of-art approaches.