멀티채널 비음수 행렬분해와 정규화된 공간 공분산 행렬을 이용한 미결정 블라인드 소스 분리

오순묵,김정한 한국음향학회 2020 韓國音響學會誌 Vol.39 No.2

This paper solves the problem in underdetermined convolutive mixture by improving the disadvantages of the multichannel nonnegative matrix factorization technique widely used in blind source separation. In conventional researches based on Spatial Covariance Matrix (SCM), each element composed of values such as power gain of single channel and correlation tends to degrade the quality of the separated sources due to high variance. In this paper, level and frequency normalization is performed to effectively cluster the estimated sources. Therefore, we propose a novel SCM and an effective distance function for cluster pairs. In this paper, the proposed SCM is used for the initialization of the spatial model and used for hierarchical agglomerative clustering in the bottom-up approach. The proposed algorithm was experimented using the ‘Signal Separation Evaluation Campaign 2008 development dataset’. As a result, the improvement in most of the performance indicators was confirmed by utilizing the ‘Blind Source Separation Eval toolbox’, an objective source separation quality verification tool, and especially the performance superiority of the typical SDR of 1 dB to 3.5 dB was verified. 본 논문은 블라인드 소스 분리 분야에서 널리 사용되는 멀티채널 비음수 행렬 분해 기법의 단점을 개선하여 미결정 복잡한 혼합 환경에서 문제를 해결한다. 공간 공분산 행렬에 기반을 둔 기존의 연구들에서, 단일 채널의 파워게 인 및 상관관계와 같은 값으로 구성된 행렬의 각 요소는 높은 분산으로 인해 분리된 소스의 품질을 저하시키는 경향이 있다. 이 논문에서는 추정된 소스들을 효과적으로 클러스터링하기 위해 레벨 및 주파수 정규화를 수행한다. 따라서 새 로운 공간 공분산 행렬 및 효과적인 클러스터 쌍별 거리함수를 제안한다. 본 논문에서는 제안된 행렬을 공간 모델의 초기화에 활용하여 공간 모델의 향상된 추정과 이를 바탕으로 상향식 접근법에서의 계층적 응집 클러스터링에 활용함 으로써 분리된 음원의 품질을 향상시켰다. 제안된 알고리즘은 ‘Signal Separation Evaluation Campaign 2008 development dataset’을 활용하여 실험을 하였다. 그 결과 객관적인 소스 분리 품질 검증 도구인 ‘Blind Source Separation Eval toolbox’를 활용하여 대부분의 성능향상지표에서의 향상을 확인하였으며, 특히 대표적인 수치인 SDR의 1 dB ~ 3.5 dB 정도의 성능우위를 검증하였다.

아동문학에 나타난 가족의 분리와 아동의 성장: 『버디가 아니라 버드』의 교육적 가치

최종녀 한국영미문학교육학회 2019 영미문학교육 Vol.23 No.2

This study analyzes the separation of families that modern society has and how children grow up in the environment. Family is important for children to develop their physical, mental and emotional growth. As time goes, the types of family are getting diverse, and it is hard to define the concept of family. The disintegration of the traditional family, divorced parents, and single parent can cause suffering or trauma to children. Thus, children’s literature tries to imply these social problems in a plot, and we expect that people would be aware of how family is important to children. This study chooses Bud, Not Buddy, a winner of Newbery Medal Award in 2000, and attempts to investigate how the protagonist, Bud, 10 year old boy, overcomes spatial, emotional separation. Also, this paper analyzes how Bud develops morality and grows self-esteem through various experiences when he goes to search for his father. In children’s literature, a parent-alternative presence emerges, giving children a sense of stability and guiding them to live in the right way. In this book, a suitcase, memories of his mother, and some stones come out as mother-alternative, and Bud relies on them mentally. Children are open minded and very flexible to adjust to difficult situations. Even though Bud couldn’t find his real father, he finds a second home where he can rest his body and mind comfortably. While facing up to the problem of separation of families, which is an unfortunate and dark side of the current society, this novel is meaningful by conveying a positive message of hope to future society through the process of the main character Bud himself growing up.

공간통계지수를 이용한 외연적 도시 확장 과정의 시공간 패턴고찰에 관한 연구: 미국 버펄로 지역을 중심으로

이경주 한국지도학회 2013 한국지도학회지 Vol.13 No.1

미국 도시에서 주거지와 상업지 간 공간적 분리는 다양한 부정적 외부효과를 파생시키는 요인 중 하나로 지적된다. 즉, 주상 토지이용 분리로 인하여 잠재적 통행발생량이 증가하고 이로 인한 교통체증, 대기오염 등의 문제가 발생하고 있다. 또한 이로부터 형성된 자동차 위주의 생활방식은 보행활동의 빈도를 감소시켜서 운동부족으로 인한 비만 및 관련 질환 발생의 간접적 요인으로 지목되기도 한다. 이 연구의 목적은 미국 도시가 보이는 외연적 확장 과정의 핵심적 특징, 즉 주거와 상업이 공간적으로 분리된 토지이용의 결과로 파생된 공간구조 형성과정을 살펴볼 수 있도록 주상 분리지수를 제안하고 이를 실증자료에 적용하여 활용성을 논의하는 것이다. 이를 위하여 미국 뉴욕 주의 이리 카운티를 대상으로 1900년부터 1995년까지 1년 단위로 주거와 상업적 토지이용의 공간적 분리과정을 분석하였다. 분석결과 이리 카운티의 경우 1920년대를 기점으로 주거와 상업이 뚜렷하게 분리된 공간구조를 형성해 왔으며, 카운티 내 교외와 도심 지역에서 각각 주거와 상업적 토지이용이 우세하게 이루어진 것을 관찰할 수 있었다. 특히 교외 지역의 고밀도 주거지를 중심으로 외연적 도시 확장으로 인한 사회적 비용절감을 목적으로 혼합적 토지이용을 유도하기 위한 정책도구의 적용이 필요할 것으로 판단된다. A spatial separation of residential and commercial land uses is often highlighted to draw a typical image of sprawled urban expansion processes. And it is argued that the spatial structure where residential land uses are dominant in urban fringe may induce excessive automobile uses and consequently, derive many negative external effects. This study aims at examining sprawled urban expansion patterns over time characterized by land use separation. For this purpose, a quantitative measure(Separation Index: SI) is devised in a spatial statistical framework. The analysis results imply that SI can be used as an effective tool for diagnosing sprawled urban structure quantified as the degree of land use separation between residential and commercial land uses. In addition, local version of SI makes it possible to identify local regions where residential suburban expansion looks problematic. For those regions, it is worth to apply smart growth strategies promoting mixed land use. The analysis framework was demonstrated using empirical data for Erie County, Buffalo, NY in 1899-1995.

Objective and Subjective Measures of Spatial Hearing in Unilateral Cochlear Implant Users with Bilateral Profound Hearing Loss

Wha Weon Jung,Jae Ho Han,Jae Hee Lee 한국청각언어재활학회 2024 Audiology and Speech Research Vol.20 No.2

Purpose: The ability to benefit from spatial separation between target and masker signals is important in multi-sound source listening environments. The goal of this study was to measure the spatial release from masking (SRM) in unilateral cochlear implant (CI) users with bilateral profound hearing loss. We also determined the relationships between the SRMs and the self-reported spatial hearing abilities. Methods: Fourteen unilateral CI users with bilateral profound hearing loss participated in this study. The target sentence was always presented to the front of the listener, and the nonfluctuating speech-shaped noise (SSN) or fluctuating speech noise was either co-located with the target (speech at 0°, noise at 0°, S0N0) or spatially separated at ± 90°. The SRM was quantified as the difference between speech recognition thresholds (SRTs) in the co-located and spatially separated conditions. The self-reported spatial hearing abilities were also measured using validated subjective questionnaires. Results: Overall, the SRTs were lower (better) with SSN than with fluctuating speech noise. When the noise was presented to the non-CI ear (speech at 0°, noise at non-CI ear, S0Nnonci), speech-in-noise recognition was the greatest due to head shadow or better-ear listening effect, resulting in the SRMs of approximately 5~6 dB regardless of noise type. When the noise was given to the CI ear (speech at 0°, noise at CI ear, S0Nci), some individuals exhibited positive SRMs (3~8 dB), while others showed negative SRMs, leading to little SRMs overall. When the SSN was given, subjects with less SRMs (less spatial separation benefits on the objective test) reported greater subjective spatial hearing difficulties. Conclusion: The spatial hearing of unilateral CI users varied by the position of the sound source. Listeners' spatial hearing abilities, which are unpredictable from clinical routine tests, need to be assessed by either objective or subjective measures.

거주지 분화에 대한 공간통계학적 접근 (Ⅱ) : 국지적 공간 분리성 측도를 이용한 탐색적 공간데이터 분석

이상일(Sang-Il Lee) 대한지리학회 2008 대한지리학회지 Vol.43 No.1

이 논문의 주된 목적은 국지적 공간 분리성 측도를 이용한 탐색적 공간데이터 분석을 통해 거주지 분화 연구에서 공간통계학적 접근이 가지는 의의를 논증하는 것이다. 탐색적 공간데이터 분석은 공간 데이터를 다양한 과학적ㆍ지도학적 시각화 방식을 통해 탐색함으로써 패턴을 발견해 내고, 의미 있는 가설을 수립하며, 더 나아가 공간 데이터에 대한 통계학적 모델을 평가하는 것을 주목적으로 한다. 이 연구는 국지 통계량에 기반한 탐색적 공간데이터 분석이 구체적인 연구 수행에서 실질적인 도움을 줄 수 있다는 믿음에 기반을 두고 진행된 것이다. 중요한 결과는 다음과 같다. 첫째, 이미 개발된 전역적 공간 분리성 측도로부터 국지적 공간 분리성 측도를 도출하였다. 둘째, 두 가지 유의성 검정을 위한 가정, 즉 총체적 랜덤화 가정과 조건적 랜덤화 가정에 기반한 가설검정 방법을 제시하였다. 셋째, 측도와 유의성 검정을 바탕으로 한 탐색적 공간데이터 분석 기법으로‘공간 분리성 산포도 지도’와‘공간 분리성 이례치 지도’를 제시하였다. 부가적으로 각 인구 집단 별 집중도에 대한 표준화 지표도 제시되었다. 넷째, 개발된 기법을 우리나라 7대 도시의 고학력 집단과 저학력 집단간 거주지 분화에 적용한 결과, 특히, 이변량 공간적 클러스터와 공간적 특이점을 확인하는 데 유용성이 있는 것으로 드러났다. The main purpose of the research is to illustrate the value of the spatial statistical approach to residential differentiation by providing a framework for exploratory spatial data analysis (ESDA) using a local spatial separation measure. ESDA aims, by utilizing a variety of statistical and cartographic visualization techniques, at seeking to detect patterns, to formulate hypotheses, and to assess statistical models for spatial data. The research is driven by a realization that ESDA based on local statistics has a great potential for substantive research. The main results are as follows. First, a local spatial separation measure is correspondingly derived from its global counterpart. Second, a set of significance testing methods based on both total and conditional randomization assumptions is provided for the local measure. Third, two mapping techniques, a ‘spatial separation scatterplot map’ and a ‘spatial separation anomaly map’, are devised for ESDA utilizing the local measure and the related significance tests. Fourth, a case study of residential differentiation between the highly educated and the least educated in major Korean metropolitan cities shows that the proposed ESDA techniques are beneficial in identifying bivariate spatial clusters and spatial outliers.

거주지 분화에 대한 공간통계학적 접근 (Ⅰ) : 공간 분리성 측도의 개발

이상일(Sang-Il Lee) 대한지리학회 2007 대한지리학회지 Vol.42 No.4

거주지 분화 현상은 도시적 삶의 공간성을 파악하는데 본질적인 요소이기 때문에 도시학 연구에서 오랫동안 주목을 받아왔다. 거주지 분화 현상에 대한 연구 과제 중의 하나가 상이한 두 집단이 얼마나 공간적으로 분리되어 있는지를 측정하는 문제이다. 이러한 측면에서 가장 널리 사용되어온 것이 상이지수(index of dissimilarity)인데, 이 지수는 거주지 분리의‘불균등성(unevenness)’은 측정할 수 있지만, 공간적‘집중도(clustering)’는 측정하지 못하는 단점을 갖고 있다. 이러한 단점을 극복하기 위해 제안되어 온‘공간적 격리 지수(spatial indices of segregation)’역시 가설검정 절차를 제시하지 못하고 최근의 공간통계학 연구 성과를 수용하지 못하는 등의 단점을 가지고 있다. 이러한 의미에서 본 논문의 주된 연구 목적은 새로운‘공간 분리성 측도(spatial separation measure)’를 개발하는 것이다. 이 공간 분리성 측도는 상이한 인구 집단이 거주 공간에 얼마나 불균등하게 분포하고 있는지에 대한 것뿐만 아니라 그러한 불균등 분포가 보여주는 공간적 의존성의 정도 까지도 측정하는 새로운 통계량이다. 주요 연구 결과는 다음과 같다. 첫째, 기존의‘공간 연관성 측도(spatial association measures)’와‘공간적 카이-스퀘어 통계량(spatial chi-square statistics)’을 통합하여 새로운 측도를 개발했으며, 일반화된 랜덤화 검정법을 적용해 측도에 대한 유의성 검정법을 제시하였다. 둘째, 개발된 측도와 유의성 검정법을 우리나라 7대 도시의 학력 집단 간 거주지 분리 현상에 적용함으로써, 연구방법론으로서의 유용성을 확인하였다. Residential differentiation is an academic theme which has been given enormous attention in urban studies. This is due to the fact that residential segregation can be seen as one of the best indicators for socio-spatial dialectics occurring on urban space. Measuring how one population group is differentiated from the other group in terms of residential space has been a focal point in the residential segregation studies. The index of dissimilarity has been the most extensively used one. Despite its popularity, however, it has been accused of inability to capture the degree of spatial clustering that unevenly distributed population groups usually display. Further, the spatial indices of segregation which have been introduced to edify the problems of the index of dissimilarity also have some drawbacks: significance testing methods have never been provided; recent advances in spatial statistics have not been extensively exploited. Thus, the main purpose of the research is to devise a spatial separation measure which is expected to gauge not only how unevenly two population groups are distributed over urban space, but also how much the uneven distributions are spatially clustered (spatial dependence). The main results are as follows. First, a new measure is developed by integrating spatial association measures and spatial chi-square statistics. A significance testing method based on the generalized randomization test is also provided. Second, a case study of residential differentiation among groups by educational attainment in major Korean metropolitan cities clearly shows the applicability of the analytical framework presented in the paper.

Benefits from Spatial Separation and Fluctuating Masker on Sentence-in-Noise Recognition in Normal-Hearing and Hearing-Impaired Listeners

Jaehoon Jung,Byoung-Duk Lim,Hyoji Shin,Jae Hee Lee 한국청각언어재활학회 2023 Audiology and Speech Research Vol.19 No.2

Purpose: This study examined the spatial separation benefit (SSB) and fluctuating masker benefit (FMB) for sentence-in-noise recognition in normal-hearing (NH) and hearing-impaired (HI) listeners. Methods: Twenty NH listeners and 10 HI listeners who were regular hearing-aid wearers participated in this study. To measure the SSB and the FMB, the Korean Matrix sentence-in-noise scores were obtained using different types of noise (steady-state speech-shaped noise, three sinusoidally amplitude-modulated noises) when the noise was colocated with the target source or was spatially separated by 30° or 60°. Results: For the NH group, the spatial separation between the target and masker was beneficial, regardless of the type of noise. Among the four types of noise, NH listeners performed poorer with speech-shaped noise than with other modulated noises, yielding a substantial improvement in speech-in-noise resulting from the masker’s fluctuation. The amount of SSB or FMB depended on the signal-to-noise ratios for the NH listeners. For HI listeners, the spatial separation was also advantageous in general, and their SSB was slightly greater in unfavorable listening conditions. However, the HI listeners hardly took advantage of the regular temporal dips of the modulated noise, even with the use of their hearing aids. Conclusion: NH listeners benefited from the fluctuation of the masker as well as the spatial separation between sound sources. A positive spatial separation benefit for HI listeners was only observed in an adverse listening condition. Regardless of the modulation rate, HI listeners received little benefit from glimpses of the target speech in the dips of the fluctuating masker. This result can be considered when planning audiological evaluation and rehabilitation for HI listeners.

Bearing Fault Diagnosis of Single-Channel Data by a 3D DCN with Bilinear LBP and Modified KPCA

Zhao Yunji,Zhou Menglin,Wang Li,Xu Xiaozhuo,Zhang Nannan 대한전기학회 2023 Journal of Electrical Engineering & Technology Vol.18 No.5

The vibration signal has the characteristics of nonlinear and non-stationary, and the distribution of fault feature information contained in it is not concentrated. In addition, the nonlinear coupling of fault-adjacent features in space is strong, resulting in poor spatial separability of fault information. At the same time, the fault diagnosis algorithm based on a convolutional neural network, cannot fully obtain the spatial distribution information of fault data due to its fixed geometric structure of convolution kernel. In order to solve the above problems, a novel fault diagnosis method for single-channel bearing fault data is proposed. First, the improved bilinear local binary pattern algorithm is used to extract time series constraint information between different points of the original fault data. Then, considering the strong nonlinear coupling of fault data adjacent features in space, this paper proposes the modified kernel principal component analysis. It obtains information on fault data in high-dimensional space by calculating the kernel space mapping matrix of different fault categories, kernelizing the sample matrix, and mapping the kernel space mapping matrix. Finally, based on this information, a 3D deformable convolution network (DCN) is introduced to obtain the spatial distribution information of fault data. DCN can adaptively adjust the shape of its own convolution kernel according to the input, which can obtain more comprehensive information and further improve the spatial separability of fault data. Experiments on CWRU and XJTU-SY both achieved 100% diagnostic accuracy, which shows the superiority of the proposed method.

매우 작은 크기의 섬광 픽셀을 사용한 배열의 평면 영상 획득을 위한 섬광체 블록 조합 검출기 개발

이승재(Seung-Jae Lee),조병두(Byungdu Jo),백철하(Cheol-Ha Baek) 한국물리학회 2024 새물리 Vol.74 No.1

Preclinical positron emission tomography (PET) detectors use very small-sized scintillation pixels to achieve high resolution. Since the scintillation pixels of this size are very small compared to the size of the photosensor pixels, the scintillation pixels located at the edges overlap in the image. To solve this problem, a light guide was applied to reduce the occurrence of overlapping by differentiating the distribution of light. In this study, a light guide made of the same material as the scintillation pixel was used to develop a detector that exhibits a much improved degree of separation compared to conventional light guides. Flood images were acquired to evaluate the degree of image separation between the edge scintillation pixels of the detector using the existing light guide and the newly developed detector. As a result of calculating the degree of separation between the two edges of the scintillation pixel image in the acquired flood imgae, the scintillator light guide showed a better degree of separation than the conventional light guide. In addition, as a result of calculating the spatial resolution of the scintillation pixel image near the edge and center, the result of using the scintillator light guide showed better spatial resolution. The superior spatial resolution of each scintillation pixel image indicates that the degree of separation between adjacent scintillation pixel images is more excellent. If the detector developed in this study is used for preclinical PET, it will be possible to improve image quality by improving overall system spatial resolution.

Application of Variable Projection Method based on Gram-Schmidt Orthogonalization in Spatial Cartesian Coordinate Transformation Model

Luyao Wang,Guolin Liu,Qiuxiang Tao,Ke Wang 대한토목학회 2019 KSCE Journal of Civil Engineering Vol.23 No.12

For the linear and nonlinear parameters that can be separated in the spatial Cartesian coordinate transformation model, we use the variable projection algorithm in this paper to represent the linear parameters with nonlinear parameters, which are transformed into least squares problems with only nonlinear parameters. We simplify the matrix of the nonlinear function by the Gram-Schmidt orthogonalization method, and combine the nonlinear least squares iterative method with the Levenberg-Marquardt (LM) algorithm to solve for the coordinate transformation parameters. Experiments are carried out by solving for the coordinate transformation parameters of the independent spatial Cartesian coordinate system and the CGCS2000 coordinate system. We compare the solution results of the four methods (parameter non-separation method, traditional variable projection method, variable projection method based on QR decomposition, and variable projection method based on Gram-Schmidt orthogonal decomposition) with respect to the calculated results, the number of iterations and the computation time. The experimental results show that the proposed method in this paper requires a lower computation time and achieves higher computational efficiency when obtaining the same solution results and with the same number of iterations.