Earticle

As an important tool to help users filter Internet information, recommender system has played a very important role wherever in academia or in industrial area. During the past years, different recommendation approaches based on the social network have been proposed with the rapid development of online social networks. Different from the traditional ones which assume all the users are independent and identically distributed, these approaches follow the intuition that a person’s implicit or explicit social network will affect his behaviors on the Web. In this paper, on the basis of the existing work, we fuse a baseline predictor model with an improved social recommendation model and propose a social recommendation algorithm based on probability matrix factorization. The experimental result shows that our method outperforms the existing approaches in accuracy.

In complicated scene, in order to solve the temporal occlusion problem of target tracking, a novel particle filter tracking algorithm based on graphical model is proposed. Graphical model is applied to particle filtering in this method. Firstly, dividing the target into several key regions, and extracting the characteristic value of each region. Then, these regions are applied to construct graphical model. In the process of target tracking by using particle filter method, graphical models can compensate for the lacking information of the occluded region. The state of the occluded part can be inferred by the graphical model. Finally, experimental results have demonstrated that the proposed tracking algorithm is effective, and it can reliably track moving target.

Although several causal models relevant to epidemiology have been proposed, a key question that has remained unanswered is why some people at high-risk for a particular disease do not develop the disease while some people at low-risk do develop it. The equivalence model, proposed herein, addresses this dilemma. The equivalence model provides a graphical description of the overall effect of risk and protective factors at the individual level. Risk factors facilitate the occurrence of the outcome (the development of disease), whereas protective factors inhibit that occurrence. The equivalence model explains how the overall effect relates to the occurrence of the outcome. When a balance exists between risk and protective factors, neither can overcome the other; therefore, the outcome will not occur. Similarly, the outcome will not occur when the units of the risk factor(s) are less than or equal to the units of the protective factor(s). In contrast, the outcome will occur when the units of the risk factor(s) are greater than the units of the protective factor(s). This model can be used to describe, in simple terms, causal inferences in complex situations with multiple known and unknown risk and protective factors. It can also justify how people with a low level of exposure to one or more risk factor(s) may be affected by a certain disease while others with a higher level of exposure to the same risk factor(s) may remain unaffected.

Spectral segmentation is a major trend in image segmentation. Specially, constrained spectral segmentation, inspired by the user-given inputs, remains its challenging task. Since it makes use of the spectrum of the affinity matrix of a given image, its overall quality depends mainly on how to design the graphical model. In this work, we propose a sparse, multi-layer graphical model, where the pixels and the over-segmented regions are the graph nodes. Here, the graph affinities are computed by using the must-link and cannot-link constraints as well as the likelihoods that each node has a specific label. They are then used to simultaneously cluster all pixels and regions into visually coherent groups across all layers in a single multi-layer framework of Normalized Cuts. Although we incorporate only the adjacent connections in the multi-layer graph, the foreground object can be efficiently extracted in the spectral framework. The experimental results demonstrate the relevance of our algorithm as compared to existing popular algorithms.

Graphical lasso is one of the most popular methods to estimate a sparse precision matrix, which is an inverse of a covariance matrix. The objective function of graphical lasso imposes an 1 A study on Bias Effect on Model Selection Criteria in Graphical Lasso Young-Geun Choi1, Seyoung Jeong2, Donghyeon Yu2* 1SK Telecom, Seoul 04539, Korea 2Department of Statistics, Inha University, Incheon 22212, Korea (Received Oct 8, 2018; Revised Nov 4, 2018; Accepted Nov 12, 2018) Abstract Graphical lasso is one of the most popular methods to estimate a sparse precision matrix, which is an inverse of a covariance matrix. The objective function of graphical lasso imposes an ℓ?-penalty on the (vectorized) precision matrix, where a tuning parameter controls the strength of the penalization. The selection of the tuning parameter is practically and theoretically important since the performance of the estimation depends on an appropriate choice of tuning parameter. While information criteria (e.g. AIC, BIC, or extended BIC) have been widely used, they require an asymptotically unbiased estimator to select optimal tuning parameter. Thus, the biasedness of the ℓ?- regularized estimate in the graphical lasso may lead to a suboptimal tuning. In this paper, we propose a two-staged bias-correction procedure for the graphical lasso, where the first stage runs the usual graphical lasso and the second stage reruns the procedure with an additional constraint that zero estimates at the first stage remain zero

본 논문은 Crandall의 탑승자 사망에 관한 모형에 Directed Graph를 응용한 것으로써 데이터는 Crandall이 사용한 미국의 1947-1981 기간의 탑승자 사망 데이터를 1993년까지 확장한 것을 사용하였다. Directed Graph Algorithm방법은 최근에 컴퓨터과학 분야에서 발전된 것을 원용한 것이다. 먼저 1947-1981 기간의 데이터를 기초로 하여 회귀분석을 통한 분석 대신에 Directed Graph Algorithm을 이용한 결과, 회귀분석을 이용했던 Crandall의 결과와는 달리 탑승자 사망은 소득수준, 자동차의 운행거리, 자동차의 안전장치 수준에 의하여 직접적으로 결정이 되는 것으로 나타났다. 자동차의 운행거리는 수득수준과 시내주행에 대한 교외주행 의 비 에 의해서 결정되는 것으로 나타났다. 이런 결과에 근거하여 3SLS(three stage least squares regression)를 이용하여 추정하고, 이러한 추정에 근거하여 1982-1993 기간을 예측했으며, Crandall의 원래의 모형의 예측력과 비교를 하여？. 예측 결과 본 모형이 MSE(mean squared error)를 기준으로한 예측력에서 훨씬 뛰어난 결과를 보였다. 더욱 중요한 것은 본고에서는 Crandall이 사용한 변수간에 기존의 계량적 방법으로는 색출이 불가능했던 잠재변수 (Latent variable)가 존재함을 구체적으로 보임으로써 회귀분석을 통한 모형화는 진정한 변수간의 관계를 반영치 못함을 보인 것이다.

Directed graphic algorithm was applied to an empirical analysis of traffic occupant fatalities based on a model by Crandall. In this paper, Crandall's data on U.S. traffic fatalities for the period 1947-1981 are focused and extended to include 1982-1993. Based on the 1947-1981 annual data, the directed graph algorithms reveal that occupant traffic deaths are directly caused by income, vehicle miles, and safety devices. Vehicle mileage is caused by income and rural driving. The estimation is conducted using three stage least squares regression. Those results show a difference between the traditional regression methodology and causal graphical analysis. It is also found that forecasts from the directed graph based model outperform forecasts from the regression-based models, in terms of mean squared forecasts error. Furthermore, it is demonstrates that there exists some latent variables between all explanatory variables and occupant deaths.

Recently, transaction data is accumulated everywhere very rapidly. Association analysis methods are usually applied to analyze transaction data, but the methods have several problems. For example, these methods can only consider one-way relations among items and cannot reflect domain knowledge into analysis process. In order to overcome defect of association analysis methods, we suggest a transaction data analysis method based on probabilistic graphical model (PGM) in this study. The method we suggest has several advantages as compared with association analysis methods. For example, this method has a high flexibility, and can give a solution to various probability problems regarding the transaction data with relationships among items.

This paper concerns 3-D graphical modeling and simulation techniques for design and control of automated equipment for construction and facility maintenance. A case study on the use of 3-D graphics techniques for developing a power plant maintenance robot is presented. By simulating equipment operation within the 3-D geometry models of the work environment the environment design was improved. The 3-D graphical models of the equipment and the work environment were further utilized for the control of the robot from a remote distance. By presenting the real-time updated equipment configuration and the work environment to the operator, the graphical model-based equipment control system helped the operator overcome the problems associated with spatial perception. The collision between the robot and the plant structures was also avoided based the real-time analysis of the dynamically updated graphical models.

오늘날 소프트웨어의 설계 및 개발과 관련된 연구들은 많은 발전을 이루고, UML과 같은 의사소통을 위한 표준 언어가 만들어졌으며 많은 사람들이 관련 이론을 수용 및 활용하고 있다. 또한, 개발 프로세스에서의 설계 및 구현과 더불어 소프트웨어의 유지 보수 단계는 매우 중요하며 이러한 소프트웨어의 유지 보수를 위한 소프트웨어 검증에 관련된 연구는 근래에 많은 주목을 받고 있다. 하지만 소프트웨어 검증의 기술 개발 수준은 설계 및 개발에 비하여 많이 미흡한 실정이다. 소프트웨어 검증은 주로 검증하기 위한 입력 데이터를 산출하여 프로그램 실행을 통해 결과를 확인하는 동적 분석에 대한 연구가 주를 이룬다. 이러한 동적 분석은 원하는 산출물의 확인을 주목적으로 하며, 결과를 표현하는 데에는 따로 정형화된 형식이 필요 없다. 하지만 소스코드를 분석하여 소프트웨어의 구조 관계와 흐름을 파악하는 정적 분석은 분석 자료를 표현하는 표현 모델이 중요하다. 현재는 정적 분석의 결과를 표현하기 위한 공통의 표현 모델이 없기 때문에 검증 과정에서의 의사소통에서 의견의 불일치의 가능성이 있고, 설계 단계에서 사용되는 표현 모델로는 정적 분석 정보의 모든 내용을 표현하는데 한계가 있다. 따라서 본 논문에서는 소프트웨어의 정적 분석 과정을 분석 4계층으로 구분하고, 각 계층마다 분석 결과를 나타내기 위한 표현 모델을 제시한다. 그리고 이 표현 모델을 활용한 소프트웨어 분석 도구의 개발을 위해, 소스 분석 데이터를 가시적으로 표현하기 위한 자료구조의 설계에 대한 내용을 다룬다.

대부분의 통계분석방법은 요약통계량에 의존하지만 그래픽적 방법을 이용하면 자료의 특성을 파악하기 쉽고 통계량만으로는 알아낼 수 없는 부분까지도 접근이 가능하다. 그래프를 통한 로지스틱회귀모형의 평가 방법으로 로그-밀도비를 통한 검토, 차원 검토, 주변모형산점도, 카이잔차산점도, CERES 그림을 알아보고 모의자료들을 통해 다양한 상황에서 그래픽적 방법들 어떠한 결과를 나타내지를 비교 검토한다.

Most statistical methods are dependent on the summary statistic. However, with graphical approaches, it is easier to identify the characteristics of the data and detect information that cannot be obtained by the summary statistic. We present various graphical methods to assess the adequacy of models in logistic regression that include checking log-density ratio, structural dimension, marginal model plot, chi-residual plot, and CERES plot. Through simulation data, we investigate and compare the results of graphical approaches under diverse conditions.

본 연구에서는 수축균열 단계를 나타낼 수 있는 도해적인 방법을 제안하였다. 우선 발생된 균열들을 균열폭의 크기 순서대로 나열하여 균열 분포를 구하였다. 다음에 균열폭을 정규화하여 0에서 1사이의 값으로 나타내었다. 마지막으로 Brooks와 Corey(1964), Fredlund와 Xing(1994), van Genuchten(1980)이 제안한바 있는 S형 곡선모델에 실험 결과를 적용시켰다. 분석 결과 van Genuchten의 식이 Brooks와 Corey식보다 정확도가 크게 높은 것으로 나타났으며, Fredlund와 Xing식보다도 높게 나타나 van Genuchten의 식을 적용하였다. 결과적으로 수축균열의 단계는 정규화 된 균열폭 분포가 3개의 직선부로 나누이는 도해적인 방법으로 나타낼 수 있었다. 제안된 방법의 적용성을 보기 위해 시료의 두께에 변화를 주며 시험을 실시하였다. 측정된 데이터를 제안된 모델에 적용하여 본 결과 높은 상관성을 보여 주었다. 따라서 수축 균열은 초기수축단계, 이차수축단계 그리고 잔류수축단계의 3단계로 모사할 수 있었다. 또한 각 단계에서의 균열 폭의 범위를 제시하였다.

The aim of this study is to present a graphical method in order to evaluate stages in shrinkage cracking. Firstly, the distribution of crack openings is established by sorting the openings of individual cracks in the soil cracking system. Secondly, it is normalized in a range of 0 to 1 to obtain the normalized crack opening distribution. Thirdly, three S-shape curve models introduced by Brooks and Corey(1964), Fredlund and Xing(1994) and van Genuchten(1980) are chosen to fit the normalized crack opening distribution using a curve fitting method. The accuracy of fitting which is described through fitting parameters by the van Genuchten equation is much higher than that by the Brooks and Corey equation and slightly higher than that by the Fredlund and Xing equation; thus the van Genuchten model is used. Finally, the stages of shrinkage cracking are graphically evaluated by drawing three separate straight lines corresponding to three linear parts of the fitted normalized crack opening distribution. The proposed method is tested with different sample thicknesses. The measured data are fitted by the selected model with the fairly high regression coefficient and small root mean square error. The results show graphically that shrinkage cracking comprises three stages; namely, primary, secondary and residual stages. Subsequently, the ranges of evaluated crack opening for each of these stages are presented.