• 한국멀티미디어학회논문지
• /
• 제19권9호
• /
• pp.1659-1668
• /
• 2016

Understanding 3D structure of scenes is of a great interest in various vision-related tasks. In this paper, we present a unified approach for estimating depth from a single monocular image. The key idea of our approach is to take advantages both of parametric learning and non-parametric sampling method. Using a parametric convolutional network, our approach learns the relation of various monocular cues, which make a coarse global prediction. We also leverage the local prediction to refine the global prediction. It is practically estimated in a non-parametric framework. The integration of local and global predictions is accomplished by concatenating the feature maps of the global prediction with those from local ones. Experimental results demonstrate that the proposed method outperforms state-of-the-art methods both qualitatively and quantitatively.

• Journal of Electrical Engineering and Technology
• /
• 제13권4호
• /
• pp.1504-1514
• /
• 2018

The objective of this study is to compare the suitability of a non-parametric and 3 parametric distributions in the characterization of prediction intervals of photovoltaic power forecasts with high confidence levels. The prediction intervals of the forecasts are calculated using a method based on recent past data similar to the target forecast input data, and on a distribution assumption for the forecast error. To compare the suitability of the distributions, prediction intervals were calculated using the proposed method and each of the 4 distributions. The calculations were done for one year of day-ahead forecasts of hourly power generation of 432 PV systems. The systems have different sizes and specifications, and are installed in different locations in Japan. The results show that, in general, the non-parametric distribution assumption for the forecast error yielded the best prediction intervals. For example, with a confidence level of 85% the use of the non-parametric distribution assumption yielded a median annual forecast error coverage of 86.9%. This result was close to the one obtained with the Laplacian distribution assumption (87.8% of coverage for the same confidence level). Contrasting with that, using a Gaussian and Hyperbolic distributions yielded median annual forecast error coverage of 89.5% and 90.5%.

• Journal of information and communication convergence engineering
• /
• 제7권4호
• /
• pp.501-506
• /
• 2009

In the H.264/ AVC video coding standard, the intra-prediction coding with various block sizes offers a considerably high improvement in coding efficiency compared to previous standards. In order to achieve this, H.264/AVC uses the Rate-distortion optimization (RDO) technique to select the best intraprediction mode for a macroblock, and it brings about the drastic increase of the computation complexity of H.264 encoder. To reduce the computation complexity and stabilize the coding performance on visual quality, this paper proposed a fast intra-prediction mode decision algorithm using non-parametric thresholds and simplified directional masks. The use of nonparametric thresholds makes the intra-coding performance not be dependent on types of video sequences and simplified directional masks reduces the compuation loads needed by the calculation of local edge information. Experiment results show that the proposed algorithm is able to reduce more than 55% of the whole encoding time with a negligible loss in PSNR and bitrates and provides the stable performance regardless types of video sequences.

• Journal of information and communication convergence engineering
• /
• 제7권2호
• /
• pp.193-198
• /
• 2009

The H.264/ AVC video coding standard supports the intra prediction with various block sizes for luma component and a 8x8 block size for chroma components. This new feature of H.264/AVC offers a considerably higher improvement in coding efficiency compared to previous compression standards. In order to achieve this, H.264/AVC uses the Rate-distortion optimization (RDO) technique to select the best intra prediction mode for each block size, and it brings about the drastic increase of the computation complexity of H.264 encoder. In this paper, a fast block size decision algorithm is proposed to reduce the computation complexity of the intra prediction in H.264/AVC. The proposed algorithm computes the smoothness based on AC and DC coefficient energy for macroblocks and compares with the nonparametric criteria which is determined by considering information on neighbor blocks already reconstructed, so that deciding the best probable block size for the intra prediction. Also, the use of non-parametric criteria makes the performance of intra-coding not be dependent on types of video sequences. The experimental results show that the proposed algorithm is able to reduce up to 30% of the whole encoding time with a negligible loss in PSNR and bitrates and provides the stable performance regardless types of video sequences.

• Journal of the Korean Statistical Society
• /
• 제28권2호
• /
• pp.225-234
• /
• 1999

We consider the following semi-parametric non-linear mixed effect regression model : y\ulcorner=f($\chi$\ulcorner;$\beta$)+$\sigma$$\mu$($\chi$\ulcorner)+$\sigma$$\varepsilon$\ulcorner,i=1,…,n,y*=f($\chi$;$\beta$)+$\sigma$$\mu$($\chi$) where y'=(y\ulcorner,…,y\ulcorner) is a vector of n observations, y* is an unobserved new random variable of interest, f($\chi$;$\beta$) represents fixed effect of known functional form containing unknown parameter vector $\beta$\ulcorner=($\beta$$_1$,…,$\beta$\ulcorner), $\mu$($\chi$) is a random function of mean zero and the known covariance function r(.,.), $\varepsilon$'=($\varepsilon$$_1$,…,$\varepsilon$\ulcorner) is the set of uncorrelated measurement errors with zero mean and unit variance and $\sigma$ is an unknown dispersion(scale) parameter. On the basis of finite-sample, small-dispersion asymptotic framework, we derive an absolute lower bound for the asymptotic mean squared errors of prediction(AMSEP) of the regular-consistent non-linear predictors of the new random variable of interest y*. Then we construct an optimal predictor of y* which attains the lower bound irrespective of types of distributions of random effect $\mu$(.) and measurement errors $\varepsilon$.

• 한국작물학회지
• /
• 제36권2호
• /
• pp.97-106
• /
• 1991

온도와 일장을 예측변수로 하는 2차원 non-par-ametric model을 개발하여, 건답직파재배에서 파종기 이동 및 단일처리 (26개품종, 4월 10일부터 2주 간격으로 8회 파종, 해지기 직전 1시간 차광)를 하여 얻은 자료로부터 출아에서 출수까지의 일평균발육속도(DVR)를 추정하였다. 또한 여기서 추정한 DVR을 이용 독립자료에 대하여 모델을 검증하였다. 1. 발육 예측정도는 온도와 일장에 대한 smoothing parameter λ$_{T}$ 와 λ$_{L}$에 따라서 단조적으로 변하였으며 예측정도를 가장 높게하는 λ$_{T}$ 와 λ$_{L}$이 존재하였다. 2. 최적 λ$_{T}$와 λ$_{L}$은 품종에 따라서 달랐으며 5～100,000의 범위내에 있었다 3. 최적 λ$_{T}$와 λ$_{L}$에서 구한 DVR을 이용하여 발육을 예측하는 경우 C.V는 품종에 따라 0.5-2.6% 였으며 기존의 함수모델들 보다 예측 정도가 높았다 4. DVR을 계산하는데 이용되지 않은 독립자료를 이용하여 11개 품종을 대상으로 출수기를 예측한 결과 예측오차는 0-3일로 추정 정도가 높았다.

• 대한지리학회지
• /
• 제51권2호
• /
• pp.219-233
• /
• 2016

수리 또는 계량적 모형을 사용하는 사회과학연구에서 분석의 초점은 종속변수와 설명변수의 관계를 밝히는 것, 즉 설명 중심의 모형(explanatory modeling)이 지금까지 주류를 이루었다. 반면 예측(prediction) 능력 제고에 초점을 맞춘 분석은 드물었다. 본 연구에서는 이론 및 가설을 검증하거나 변수 간의 관계를 밝히는 설명 중심의 모형이 아니라 신규 관찰치에 대한 예측 오차를 줄이는, 예측 중심의 비모수 모형(non-parametric model)을 검토하였다. 서울시 강남구를 사례지역으로 선정한 후, 2011년부터 2014년까지 신고된 단독주택 실거래가를 기초자료로 하여 주택가격을 추정하였다. 적용한 비모수 모형은 기계학습 분야에서 제시된 일반가산모형(generalized additive model), 랜덤 포리스트, MARS(multivariate adaptive regression splines), SVM(support vector machines) 등이며 비교적 최근에 개발된 MARS나 SVM의 예측력이 뛰어남을 확인할 수 있었다. 마지막으로 이러한 비모수 모형에 공간적 자기상관성을 추가적으로 반영한 결과, 모형의 가격 예측력이 보다 개선되었음을 알 수 있었다. 본 연구를 계기로 그간 모수 모형에 집중되었던 부동산 가격추정 방법론이 비모수 모형으로 확대 및 다양화되기를 기대한다.

• 한국지구과학회지
• /
• 제26권2호
• /
• pp.114-128
• /
• 2005

Fuzzy information representation of multi-source spatial data is applied to landslide hazard mapping. Information representation based on frequency ratio and non-parametric density estimation is used to construct fuzzy membership functions. Of particular interest is the representation of continuous data for preventing loss of information. The non-parametric density estimation method applied here is a Parzen window estimation that can directly use continuous data without any categorization procedure. The effect of the new continuous data representation method on the final integrated result is evaluated by a validation procedure. To illustrate the proposed scheme, a case study from Jangheung, Korea for landslide hazard mapping is presented. Analysis of the results indicates that the proposed methodology considerably improves prediction capabilities, as compared with the case in traditional continuous data representation.

• 대한산업공학회지
• /
• 제41권4호
• /
• pp.408-413
• /
• 2015

The quality prediction of the semiconductor industry has been widely recognized as important and critical for quality improvement and productivity enhancement. The main objective of this paper is to predict the final quality of semiconductor chips based on fail bit count information obtained from probe tests. Our proposed method consists of solving the data imbalance problem, non-parametric variable selection, and adjusting the parameters of the model. We demonstrate the usefulness and applicability of the proposed procedure using a real data from a semiconductor manufacturing.

• 응용통계연구
• /
• 제30권5호
• /
• pp.701-717
• /
• 2017

급속한 고령화로 인하여 미래의 인구와 인구구조에 관해 사회와 정부의 관심이 증가하고 있으며 우리나라의 사망률은 감소하고 있으나 감소폭은 변동적이다. 본 연구에서는 이를 고려할 수 있는 모형을 살펴보고자 LC 모형, LM 모형, BMS 모형 그리고 비모수평활 기법이 적용된 FDM과 Coherent FDM을 비교 분석하여 연령별 사망률과 기대수명 예측의 정확성 측면에서 남녀 사망률 개선 추이를 예측하는데 적합한 모형을 살펴보았다. 또한 우리나라 사망률 예측에 비모수 기법의 활용 가능성을 검토하였다. 분석 결과 최근 자료의 추세를 잘 반영하는 비모수기법을 활용한 인구통계모델인 FDM과 Coherent FDM의 예측력이 우수함을 알 수 있었다. 결과적으로 FDM과 Coherent FDM은 적합이 뛰어나고, 미래에 변화가 크지 않다면 예측력 또한 우수하다 볼 수 있을 것이다.