• 대한방사성의약품학회지
• /
• 제3권2호
• /
• pp.65-71
• /
• 2017

To assess the effects of filter and reconstruction of Cu-64 PET data on Siemens scanner, the various reconstruction algorithm with various filters were assessed in terms of spatial resolution, non-uniformity (NU), recovery coefficient (RC), and spillover ratio (SOR). Image reconstruction was performed using filtered backprojection (FBP), 2D ordered subset expectation maximization (OSEM), 3D reprojection algorithm (3DRP), and maximum a posteriori algorithms (MAP). For the FBP reconstruction, ramp, butterworth, hamming, hanning, or parzen filters were used. Attenuation or scatter correction were performed to assess the effect of attenuation and scatter correction. Regarding spatial resolution, highest achievable volumetric resolution was $3.08mm^3$ at the center of FOV when MAP (${\beta}=0.1$) reconstruction method was used. SOR was below 4% for FBP when ramp, Hamming, Hanning, or Shepp-logan filter were used. The lowest NU (highest uniform) after attenuation & scatter correction was 5.39% when FBP (parzen filter) was used. Regarding RC, 0.9 < RC < 1.1 was obtained when OSEM (iteration: 10) was used when attenuation and scatter correction were applied. In this study, image quality of Cu-64 on Siemens Inveon PET was investigated. This data will helpful for the quantification of Cu-64 PET data.

• 한국통신학회논문지
• /
• 제24권10B호
• /
• pp.1902-1911
• /
• 1999

본 논문에서는 객체들의 공간적 특성이 반영된 시각적인 특징벡터를 이용한 내용기반 영상검색 알고리즘을 제안한다. 제안된 검색 기법은 여러 색상으로 이루어진 객체들을 표현하기 위하여 가우시안 혼성 모델을 적용하여 모델의 최대유사 파라미터는 EM 알고리즘을 사용하여 추정한다. GMM을 기반으로 하여 분할된 각 객체들로부터 Fourier descriptor의 색상 히스토그램을 사용하여 모양과 색상 특징을 추출하게 된다. 영상 검색은 두 단계로 구성되는데 첫 단계에서는 공간적인 모양 특성을 추출하여 모양이 유사한 객체들을 후보 영상으로 압축하게 되며 마지막으로 객체의 색상 히스토그램에 의하여 검색이 수행된다. 실험 결과 제안된 알고리즘은 분할된 객체의 공간적, 시각적 특징을 이용하여 효율적으로 검색을 수행할 수 있음을 보여준다.

• 전자공학회논문지
• /
• 제50권2호
• /
• pp.185-192
• /
• 2013

본 논문에서는 MFCC를 이용한 GMM 기반의 음성과 혼합 신호 분류 알고리즘을 MPEG의 표준 코덱인 USAC에 적용하였다. 효과적인 패턴 인식을 위해 GMM을 이용하였고, EM알고리즘을 사용하여 최적의 GMM 파라미터를 추출하였다. 제안하는 분류 알고리즘은 두 가지 중요한 부분으로 나뉜다. 첫째는 GMM을 통해 최적의 파라미터를 추출하는 것 이고, 두 번째는 MFCC 값을 이용한 패턴인식을 통해 음성/혼합 신호를 분류하였다. 제안된 알고리즘의 성능을 평가한 결과 MFCC를 이용한 GMM 기반의 제안된 방법이 기존 USAC의 방법보다 우수한 음성/혼합 신호 분류 성능을 보였다.

• Genomics & Informatics
• /
• 제5권2호
• /
• pp.61-67
• /
• 2007

The allele frequencies of markers as well as linkage disequilibrium (LD) can be changed in cases due to the LD between markers and the disease allele, exhibiting spurious associations of markers. To identify the true association, classical statistical tests for dealing with confounders have been applied to draw a conclusion as to whether the association of variants comes from LD with the known disease allele. However, a more direct test considering LD using estimated haplotype frequencies may be more efficient. The null hypothesis is that the different allele frequencies of a variant between cases and controls come solely from the increased disease allele frequency and the LD relationship with the disease allele. The haplotype frequencies of controls are estimated using the expectation maximization (EM) algorithm from the genotype data. The estimated frequencies are applied to calculate the expected haplotype frequencies in cases corresponding to the increase or decrease of the causative or protective alleles. The suggested method was applied to previously published data, and several APOE variants showed association with Alzheimer's disease independent from the APOE ${\varepsilon}4$ variant, rs429358, regardless of LD showing significant simulated p-values. The test results support the possibility that there may be more than one common disease variant in a locus.

• 제어로봇시스템학회논문지
• /
• 제19권6호
• /
• pp.513-519
• /
• 2013

The current vision-based approaches for emotion recognition, such as facial expression analysis, have many technical limitations in real circumstances, and are not suitable for applications that use them solely in practical environments. In this paper, we propose an approach for emotion recognition by combining extrinsic representations and intrinsic activities among the natural responses of humans which are given specific imuli for inducing emotional states. The intrinsic activities can be used to compensate the uncertainty of extrinsic representations of emotional states. This combination is done by using PRMs (Probabilistic Relational Models) which are extent version of bayesian networks and are learned by greedy-search algorithms and expectation-maximization algorithms. Previous research of facial expression-related extrinsic emotion features and physiological signal-based intrinsic emotion features are combined into the attributes of the PRMs in the emotion recognition domain. The maximum likelihood estimation with the given dependency structure and estimated parameter set is used to classify the label of the target emotional states.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2008년도 한국컴퓨터종합학술대회논문집 Vol.35 No.1 (C)
• /
• pp.482-486
• /
• 2008

본 논문에서는 하드 디스크 드라이브(Hard Disk Drive, HDD) 생산 공정 과정에서 발생할 수 있는 불량 HDD의 결함 분포에 대해서 패턴을 자동으로 분류해주는 기법을 제시한다. 이를 위해서 표준 패턴 클래스로 분류되어 있는 불량 HDD의 각 클래스의 확률 모델을 GMM(Gaussian Mixture Model)로 가정한다. 실험은 전문가에 의해 분류된 실제 HDD 결함 분포로부터 5가지의 특징 값들을 추출한 후, 결함 분포의 클래스를 표현할 수 있는 GMM의 파라미터(Parameter)를 학습한다. 각 모델의 파라미터를 추정하기 위해 EM(Expectation Maximization) 알고리즘을 사용한다. 학습된 GMM의 분류 테스트는 학습에 사용되지 않은 HDD 결함 분포에서 5가지의 특징 값을 입력 값으로 추정된 모델들의 파라미터 값에 의해 사후 확률을 구한다. 계산된 확률 값 중 가장 큰 값을 갖는 모델의 클래스를 표준 패턴 클래스로 분류한다. 그 결과 제시된 GMM을 이용한 HDD의 패턴 분류의 결과 96.1%의 정답률을 보여준다.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제13권3호
• /
• pp.186-199
• /
• 2013

We pose pattern classification as a density estimation problem where we consider mixtures of generative models under partially labeled data setups. Unlike traditional approaches that estimate density everywhere in data space, we focus on the density along the decision boundary that can yield more discriminative models with superior classification performance. We extend our earlier work on the recursive estimation method for discriminative mixture models to semi-supervised learning setups where some of the data points lack class labels. Our model exploits the mixture structure in the functional gradient framework: it searches for the base mixture component model in a greedy fashion, maximizing the conditional class likelihoods for the labeled data and at the same time minimizing the uncertainty of class label prediction for unlabeled data points. The objective can be effectively imposed as individual mixture component learning on weighted data, hence our mixture learning typically becomes highly efficient for popular base generative models like Gaussians or hidden Markov models. Moreover, apart from the expectation-maximization algorithm, the proposed recursive estimation has several advantages including the lack of need for a pre-determined mixture order and robustness to the choice of initial parameters. We demonstrate the benefits of the proposed approach on a comprehensive set of evaluations consisting of diverse time-series classification problems in semi-supervised scenarios.

• Communications for Statistical Applications and Methods
• /
• 제23권2호
• /
• pp.131-146
• /
• 2016

In research on behavioral studies, significant attention has been paid to the stage-sequential process for longitudinal data. Latent class profile analysis (LCPA) is an useful method to study sequential patterns of the behavioral development by the two-step identification process: identifying a small number of latent classes at each measurement occasion and two or more homogeneous subgroups in which individuals exhibit a similar sequence of latent class membership over time. Maximum likelihood (ML) estimates for LCPA are easily obtained by expectation-maximization (EM) algorithm, and Bayesian inference can be implemented via Markov chain Monte Carlo (MCMC). However, unusual properties in the likelihood of LCPA can cause difficulties in ML and Bayesian inference as well as estimation in small samples. This article describes and addresses erratic problems that involve conventional ML and Bayesian estimates for LCPA with small samples. We argue that these problems can be alleviated with a small amount of prior input. This study evaluates the performance of likelihood and MCMC-based estimates with the proposed prior in drawing inference over repeated sampling. Our simulation shows that estimates from the proposed methods perform better than those from the conventional ML and Bayesian method.

• 한국측량학회지
• /
• 제34권2호
• /
• pp.195-206
• /
• 2016

The remote sensing technique using SAR data have many advantages when applied to the disaster site due to its wide coverage and all-weather acquisition availability. Although a single-pol (polarimetric) SAR image cannot represent the land surface better than a quad-pol SAR image can, single-pol SAR data are worth using for disaster-induced change detection. In this paper, an automatic change detection method based on a mixture of GGDs (generalized Gaussian distribution) is proposed, and usability of the textural features and intensity is evaluated by using the proposed method. Three ALOS/PALSAR images were used in the experiments, and the study site was Norita City, which was affected by the 2011 Tohoku earthquake. The experiment results showed that the proposed automatic change detection method is practical for disaster sites where the large areas change. The intensity information is useful for detecting disaster-induced changes with a 68.3% g-mean, but the texture information is not. The autocorrelation and correlation show the interesting implication that they tend not to extract agricultural areas in the change detection map. Therefore, the final tsunami-induced change map is produced by the combination of three maps: one is derived from the intensity information and used as an initial map, and the others are derived from the textural information and used as auxiliary data.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제6권9호
• /
• pp.2351-2369
• /
• 2012

This paper describes interpolation method of motion field in the Wyner-Ziv video coding (WZVC) based on Expectation-Maximization (EM) algorithm. In the EM algorithm, the estimated motion field distribution is calculated on a block-by-block basis. Each pixel in the block shares similar probability distribution, producing an undesired blocking artefact on the pixel-based motion field. The proposed interpolation techniques are Bicubic and Lanczos which successively use 16 and 32 neighborhood probability distributions of block-based motion field for one pixel in k-by-k block on pixel-based motion field. EM-based WZVC codec updates the estimated probability distribution on block-based motion field, and interpolates it to pixel resolution. This is required to generate higher-quality soft side information (SI) such that the decoding algorithm is able to make syndrome estimation more quickly. Our experiments showed that the proposed interpolation methods have the capability to reduce EM-based WZVC decoding complexity with small increment of bit rate.