• Journal of Computing Science and Engineering
• /
• 제7권2호
• /
• pp.99-111
• /
• 2013

Approximating a divergence between two probability distributions from their samples is a fundamental challenge in statistics, information theory, and machine learning. A divergence approximator can be used for various purposes, such as two-sample homogeneity testing, change-point detection, and class-balance estimation. Furthermore, an approximator of a divergence between the joint distribution and the product of marginals can be used for independence testing, which has a wide range of applications, including feature selection and extraction, clustering, object matching, independent component analysis, and causal direction estimation. In this paper, we review recent advances in divergence approximation. Our emphasis is that directly approximating the divergence without estimating probability distributions is more sensible than a naive two-step approach of first estimating probability distributions and then approximating the divergence. Furthermore, despite the overwhelming popularity of the Kullback-Leibler divergence as a divergence measure, we argue that alternatives such as the Pearson divergence, the relative Pearson divergence, and the $L^2$-distance are more useful in practice because of their computationally efficient approximability, high numerical stability, and superior robustness against outliers.

• 한국안광학회지
• /
• 제14권4호
• /
• pp.53-57
• /
• 2009

목적: 각각의 굴절이상도에 따른 폭주근점(N.P.C), 근거리 수평사위, 근거리 폭주 개산여력을 알아보고자 하는 연구이다. 방법: 안과적 수술이나 질환이 없는 대학생 39명(평균 22.7세)을 대상으로 굴절이상도, 폭주근점, 근거리 수평사위, 근거리 폭주 및 개산여력을 측정하였다. 결과: 검사 결과를 경도, 중도, 고도 굴절이상에 따라 모건분류 정상 기대값과 비교하여 분류하였다. 폭주근점(NPC)은 경도, 중등도 모두 8 cm 이내로 검출된 반면 강도 굴절이상에서는 9.64 cm가 검출되었다. 경도 및 중등도 굴절이상도에서 근거리 수평사위에 의한 분류를 하였을 경우 내사위의 경우 근거리 개산여력이 가장 적었으며, 반면 폭주여력에서는 강한 외사위 그룹이 적게 검출되었다. 결론: 폭주여력은 내사위에서 크고 외사위에서 작은 반면, 개산여력은 내사위에서 적고 외사위에서 크게 나타났다.

• Journal of applied mathematics & informatics
• /
• 제26권3_4호
• /
• pp.733-738
• /
• 2008

Using an information of dimensions of divergence points, we give full information of dimensions of the completely decomposed class of the lower(upper) distribution sets of a self-similar Cantor set. Further using a relationship between the distribution sets and the subsets generated by the lower(upper) local dimensions of a self-similar measure, we give full information of dimensions of the subsets by the local dimensions.

• 한국안광학회지
• /
• 제5권1호
• /
• pp.101-105
• /
• 2000

본 연구는 한국인을 대상으로 주시점의 위치에 따른 시 표적이 뇌파에 미치는 영향을 조사한 것이다. 시유발 전위 장치는 BIO-Pag를 이용하였으며, 검사 결과는 586 컴퓨터에 입력하여 분석하였다. 조도는 500lux이여 시표는 직경이 3 cm인 붉은 광점을 사용하였다. 수렴과 개산에 따른 결과는 다음과 같다. 1. 시자극에 관계하는 시피질의 뇌파의 출현 빈도는 델타파, 베타파, 세타파, 그리고 알파파의 순으로 나타났다. 2. 수렴 상태는 개산에 비해서 속파 성분이 많이 출현되었다. 3. 수렴은 개산에 비해서 베타파와 알파파가 많아 나타났다. 4. 수렴과 개산에서 진폭에 대한 히스토 그램은 비가우시안 모양이다. 5. 뇌파의 진폭에 대한 위상 분석은 거의 선상을 나타내었다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1998년도 제13차 학술회의논문집
• /
• pp.232-237
• /
• 1998

This paper describes a new method for estimating time-to-collision which exhibits high tolerance to noise contained in camera images. Time to collision (TTC) is one of the most important parameters available from a camera attached to a mobile machine. TTC indirectly stands far the translation speed of the camera and is usually calculated either from successive images or optical flow by using intimate relationship between TTC and flow divergence. In most cases, however, it is not easy to get accurate optical flow, which makes it difficult to calculate TTC. In this study it is proved that if the target has a smooth surface, the average of divergence over any point-symmetric region on the image is equal to the divergence of the center of the region. It means that required divergence can be calculated by integrating optical flow vectors over a symmetric region. It is expected that in the process of the integration, accidental noise is canceled if they are independent of optical flow and the motion of the camera. Experimental results show that TTC can be estimated regardless of the surface condition. It is also shown that influence of noise is eliminated as the area of integration increases.

• 한국안광학회지
• /
• 제10권3호
• /
• pp.205-213
• /
• 2005

교정시력이 1.0 이상인 21세에서 30세 사이인 대학생 90명(남25명, 여65명)을 대상으로 포롭터를 이용하여 양안시 검사를 하였다. 원거리에서 허성상대 폭주력(외전)의 기대치(흐린점/분리점/회복점)내에 포함되는 대상의 비율은 78%/61%/67%로 나타났고, 개산과다는 9%/31%/23%, 개산부족은 13%/8%/9%이었다. 원거리에서 실성상대폭주력(내전)은 20%/46%/39%, 폭주과다는 22%/14%/16%, 폭주부족은 35%/40%/45%를 각각 보였다. AC/A비 측정에서 42명이 정상이었고 38명은 낮음, 나머지는 높음을 보였으며 낮은 AC/A비는 조절관계에서 낮은 이향운동의 결과로 볼 수 있다. 근거리에서 허성상대 폭주력 기대치는 26%/29%/44%로 나타났고, 개산과다는 61%/33%/24%, 개산부족은 3%138%/32%이었으며 실성상대 폭주력의 기대치내에 포함되는 대상의 비율은 33%/40%/31%, 폭주과다는 61%/23%/42%, 폭주부족은 6%/37%/27%로 각각 나타났다. 폭주근점 검사에서 58%는 8cm 이내로 기대치에 포함되었다. 허성상대조절력(NRA)과 실성상대조절력(PRA) 검사 결과 기대지내에 포함되는 대상의 비율은 각각 41%와 33%였다. 그리고 조절부족은 33%와 43% 및 조절과다는 26%와 24%를 각각 나타냈다. 조절과다는 NRA의 낮은값과 관련되었고 초절부족과 폭주과다는 NRA의 높은 값과 관련 있고, 조절과다이면서 폭주과다인 기능 이상은 NRA의 높은 값과 관련 있는 것으로 나타났다.

• Journal of the Korean Statistical Society
• /
• 제30권2호
• /
• pp.247-263
• /
• 2001

This paper concerns a problem of classification based on training dta. A framework of information geometry is given to elucidate the characteristics of discriminant functions including logistic discrimination and AdaBoost. We discuss a class of loss functions from a unified viewpoint.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2004년도 추계학술대회
• /
• pp.1546-1551
• /
• 2004

The laser Doppler technique is well-established as a velocity measurement technique of high precision for flow velocity. Recently, the laser Doppler technique has also been used to measure acceleration of fluid particles. Acceleration is interesting from a fluid mechanics point of view, since the Navier Stokes equations, specifically the left-hand-side, are formulated in terms of fluid acceleration. Further, there are several avenues to estimating the dissipation rate using the acceleration. However such measurements place additional demands on the design of the optical system; in particular fringe non-uniformity must be held below about 0.0001 to avoid systematic errors. Relations expressing fringe divergence as a function of the optical parameters of the system have been given in the literature; however, direct use of these formulae to minimize fringe divergence lead either to very large measurement volumes or to extremely high intersection angles. This dilemma can be resolved by using an off-axis receiving arrangement, in which the measurement volume is truncated by a pinhole in front of the detection plane. In the present study an optical design study is performed for optimizing laser Doppler systems for fluid acceleration measurements. This is followed by laboratory validation using a round free jet and a stagnation flow, two flows in which either fluid acceleration has been previously measured or in which the acceleration is known analytically. A 90 degree off-axis receiving angle is used with a pinhole or a slit.

• Nuclear Engineering and Technology
• /
• 제55권4호
• /
• pp.1350-1364
• /
• 2023

2D/1D coupling method is an important neutron transport calculation method due to its high accuracy and relatively low computation cost. However, 2D/1D coupling method may diverge especially in small axial mesh size. To analyze the convergence behavior of 2D/1D coupling method, a Fourier analysis for k-eigenvalue neutron transport problems is implemented. The analysis results present the divergence problem of 2D/1D coupling method in small axial mesh size. Several common attempts are made to solve the divergence problem, which are to increase the number of inner iterations of the 2D or 1D calculation, and two times 1D calculations per outer iteration. However, these attempts only could improve the convergence rate but cannot deal with the divergence problem of 2D/1D coupling method thoroughly. Moreover, the choice of axial solvers, such as DGFEM SN and traditional SN, and its effect on the convergence behavior are also discussed. The results show that the choice of axial solver is a key point for the convergence of 2D/1D method. The DGFEM SN based 2D/1D method could converge within a wide range of optical thickness region, which is superior to that of traditional SN method.

• Journal of the Korean Data and Information Science Society
• /
• 제17권2호
• /
• pp.543-557
• /
• 2006

This paper is concerned with the applicability of the chi-square approximation to the six disparity statistics: the Pearson chi-square, the generalized likelihood ratio, the power divergence, the blended weight chi-square, the blended weight Hellinger distance, and the negative exponential disparity statistic. Three dimensional contingency tables of small and moderate sample sizes are generated to be fitted to all possible hierarchical log-linear models: the completely independent model, the conditionally independent model, the partial association models, and the model with one variable independent of the other two. For models with direct solutions of expected cell counts, point estimates and confidence intervals of the 90 and 95 percentage points of six statistics are explored. For model without direct solutions, the empirical significant levels and the empirical powers of six statistics to test the significance of the three factor interaction are computed and compared.