• Journal of the Korean Data and Information Science Society
• /
• 제23권3호
• /
• pp.579-585
• /
• 2012

Many different semi-supervised learning algorithms have been proposed for use wit unlabeled data. However, most of them focus on classification problems. In this paper we propose a semi-supervised regression algorithm called the semi-supervised local constant estimator (SSLCE), based on the local constant estimator (LCE), and reveal the asymptotic properties of SSLCE. We also show that the SSLCE has a faster convergence rate than that of the LCE when a well chosen weighting factor is employed. Our experiment with synthetic data shows that the SSLCE can improve performance with unlabeled data, and we recommend its use with the proper size of unlabeled data.

• 한국지능시스템학회논문지
• /
• 제7권2호
• /
• pp.96-105
• /
• 1997

본 논문에서는 복잡한 비선형 시스템의 모델동정을 위해 퍼지모델링의 새로운 방법이 제안된다. 제안된 FPNN모델링은 공정시스템의 입출력 데이터로부터 GMDH방법과 퍼지구현규칙을 이용하여 시스템의 구조와 파라미터 동정을 구현한다. 퍼지구현규칙의 전반부 구조와 파라미터 동정을 위하여 GMDH 방법과 희귀다항식 퍼지추론 방법이 사용되고 최적 후반부 파라미터 동정을 위하여 최소자승법이 사용된다. 가스로 시계열데이타 및 하수처리시스템의 활성화의 공정 데이터가 제안한 FPNN 모델링의 성능을 평가하기 위해 상용된다. 제안된 방법이 기존의 다른 논문과 비교하여 더 높은 정확도를 가진 지능형 모델을 생성함을 보인다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2003년도 하계학술대회 논문집 D
• /
• pp.2519-2521
• /
• 2003

In this paper, we propose Self-Organizing Fuzzy Polynomial Neural Networks(SOFPNN) architecture for optimal model identification and discuss a comprehensive design methodology supporting its development. It is shown that this network exhibits a dynamic structure as the number of its layers as well as the number of nodes in each layer of the SOFPNN are not predetermined (as this is the case in a popular topology of a multilayer perceptron). As the form of the conclusion part of the rules, especially the regression polynomial uses several types of high-order polynomials such as linear, quadratic, and modified quadratic. As the premise part of the rules, both triangular and Gaussian-like membership function are studied and the number of the premise input variables used in the rules depends on that of the inputs of its node in each layer. We introduce two kinds of SOFPNN architectures, that is, the basic and modified one with both the generic and the advanced type. The superiority and effectiveness of the proposed SOFPNN architecture is demonstrated through nonlinear function numerical example.

• 지능정보연구
• /
• 제28권3호
• /
• pp.105-117
• /
• 2022

본 연구는 다항회귀분석을 통해 장기금리와 단기금리의 차이인 금리 스프레드와 주식 수익률 간 영향을 분석한다. 기존 연구들은 미국시장을 중심으로 금리 스프레드를 통한 경기를 예측에 초점을 맞추어 진행되었다. 선행 연구들은 장단기금리의 기간을 조절하고 선행정도를 분석하며 금리 스프레드를 경기예측 선행지표로 검증했다. 국내에서도 2006년 경기종합지수 제 7차 개편 이후 금리스프레드를 경기 선행지수 구성항목에 포함하였으며 현재까지도 활용하고 있다. 그럼에도 불구하고 국내 주식시장에서 금리스프레드와 산업별 주식 수익률에 대한 연구는 부족하다. 때문에 본 연구에서는 국내주식시장을 대상으로 금리스프레드와 산업별 주식 수익률은 분석했다. 회귀분석을 통해 인과관계가 높은 장단기 금리를 선정하고 선행기간 및 산업별 상관관계를 파악했다. 연구 과정에서 단순 선형회귀 분석(Simple Linear Regression)의 한계를 극복하기 위해 다항 회귀분석(Polynomial Linear Regression)을 활용해 설명력을 높였다. 분석 결과 6개월 선행하여 무보증 3년 회사채(AA-) 수익률과 콜금리 수익률의 차이 금리스프레드로 사용했을 때 높은 인과를 확인하였으며 산업별 주식수익률을 분석한 결과 해당 금리 스프레드와 자동차산업의 수익률의 관계가 가장 밀접함을 확인했다. 본 연구를 통해 국내에서 금리 스프레드가 경기예측뿐만 아니라 주식수익률과도 인과관계가 있음을 확인한 것에 의의가 있다. 금리스프레드만 사용하여 주식 가격을 예측하는 것에는 한계가 있을 수 있으나 다양한 요인들과 적절히 활용할 경우 강력한 팩터로 역할을 할 것이라 기대한다.

• Journal of Electrical Engineering and Technology
• /
• 제9권6호
• /
• pp.2098-2106
• /
• 2014

In the area of clustering, there are numerous approaches to construct clusters in the input space. For regression problem, when forming clusters being a part of the overall model, the relationships between the input space and the output space are essential and have to be taken into consideration. Conditional Fuzzy C-Means (c-FCM) clustering offers an opportunity to analyze the structure in the input space with the mechanism of supervision implied by the distribution of data present in the output space. However, like other clustering methods, c-FCM focuses on the distribution of the data. In this paper, we introduce a new method, which by making use of the ambiguity index focuses on the boundaries of the clusters whose determination is essential to the quality of the ensuing classification procedures. The introduced design is illustrated with the aid of numeric examples that provide a detailed insight into the performance of the fuzzy classifiers and quantify several essentials design aspects.

• 말소리와 음성과학
• /
• 제9권3호
• /
• pp.17-23
• /
• 2017

The present study investigates the effect of Accentual Phrase on F0 using a subset of large-scale corpus of Seoul Korean. Four syllable words which were neither preceded nor followed by silent pauses were presumed to be canonical exemplars of Accentual Phrases in Korean. These four syllable words were extracted from female speakers' speech samples. Growth curve analyses, combination of regression and polynomial curve fitting, were applied to the four syllable words. Four syllable words were divided into four groups depending on the categorical status of the initial segment: voiceless obstruents, voiced obstruents, sonorants, and vowels. Results of growth curve analyses indicate that initial segment types have an effect on the F0 (in semitone) in the nucleus of the initial syllable, and the cubic polynomial term revealed that some of the medial low tones in the 4 syllable words may be guided by the principle of contrast maximization, while others may be governed by the principle of ease of articulation.

• Journal of the Korean Data and Information Science Society
• /
• 제15권3호
• /
• pp.565-574
• /
• 2004

Classification tree algorithms including as CART by Brieman et al.(1984) in some aspects, recursively partition the data space with the aim of making the distribution of the class variable as pure as within each partition and consist of several steps. SUPPORT(smoothed and unsmoothed piecewise-polynomial regression trees) method of Chaudhuri et al(1994), a weighted averaging technique is used to combine piecewise polynomial fits into a smooth one. We focus on applying SUPPORT to a binary class variable. Logistic model is considered in the caculation techniques and the results are shown good classification rates compared with other methods as CART, QUEST, and CHAID.

• Geomechanics and Engineering
• /
• 제23권6호
• /
• pp.547-560
• /
• 2020

This paper studies the dynamic foundation-soil-foundation interaction for two square rigid foundations embedded in a viscoelastic soil layer. The vibrations come from only one rigid foundation placed in the soil layer and subjected to harmonic loads of translation, rocking, and torsion. The required dynamic response of rigid surface foundations constitutes the solution of the wave equations obtained by taking account of the conditions of interaction. The solution is formulated using the frequency domain Boundary Element Method (BEM) in conjunction with the Kausel-Peek Green's function for a layered stratum, with the aid of the Thin Layer Method (TLM), to study the dynamic interaction between adjacent foundations. This approach allows the establishment of a mathematical model that enables us to determine the dynamic displacements amplitude of adjacent foundations according to their different separations, the depth of the substratum, foundations masss, foundations embedded, and the frequencies of excitation. This paper attempts to introduce an approach based on a polynomial mathematical tool conducted from several results of numerical methods (BEM-TLM) so that practicing civil engineers can evaluation the dynamic foundations displacements more easy.

• 한국농공학회논문집
• /
• 제62권1호
• /
• pp.61-70
• /
• 2020

To estimate the ventilation volume of mechanically ventilated swine farms, various regression models were applied, and errors were compared to select the regression model that can best simulate actual data. Linear regression, linear spline, polynomial regression (degrees 2 and 3), logistic curve, generalized additive model (GAM), and gompertz curve were compared. Overfitting models were excluded even when the error rate was small. The evaluation criteria were root mean square error (RMSE) and mean absolute percentage error (MAPE). The evaluation results indicated that degree 3 exhibited the lowest error rate; however, an overestimation contradiction was observed in a certain section. The logistic curve was the most stable and superior to all the models. In the estimation of ventilation volume by all of the models, the estimated ventilation volume of the logistic curve was the smallest except for the model with a large error rate and the overestimated model.

• 한국환경과학회지
• /
• 제23권3호
• /
• pp.359-368
• /
• 2014

The aim of our research was to apply experimental design methodology in the optimization of N, N-Dimethyl-4-nitrosoaniline (RNO, which is indictor of OH radical formation) degradation using gas mixing-circulation plasma process. The reaction was mathematically described as a function of four independent variables [voltage ($X_1$), gas flow rate ($X_2$), liquid flow rate ($X_3$) and time ($X_4$)] being modeled by the use of the central composite design (CCD). RNO removal efficiency was evaluated using a second-order polynomial multiple regression model. Analysis of variance (ANOVA) showed a high coefficient of determination ($R^2$) value of 0.9111, thus ensuring a satisfactory adjustment of the second-order polynomial multiple regression model with the experimental data. The application of response surface methodology (RSM) yielded the following regression equation, which is an empirical relationship between the RNO removal efficiency and independent variables in a coded unit: RNO removal efficiency (%) = $77.71+10.04X_1+10.72X_2+1.78X_3+17.66X_4+5.91X_1X_2+3.64X_2X_3-8.72X_2X_4-7.80X{_1}^2-6.49X{_2}^2-5.67X{_4}^2$. Maximum RNO removal efficiency was predicted and experimentally validated. The optimum voltage, air flow rate, liquid flow rate and time were obtained for the highest desirability at 117.99 V, 4.88 L/min, 6.27 L/min and 24.65 min, respectively. Under optimal value of process parameters, high removal(> 97 %) was obtained for RNO.