• Proceedings of the Korea Water Resources Association Conference
• /
• 2015.05a
• /
• pp.414-414
• /
• 2015

최근 지구온난화로 인한 기상변동성 증가로 인해 극한기후현상의 발생빈도가 점차 증가하고 있으며 유역단위의 수자원을 효율적으로 운영하는데 문제점을 해소하고자 다양한 측면에서 체계적인 수자원 운영을 위한 연구가 이루어지고 있다. 수공구조물을 설계하는데 있어서 가장 일반적인 가정 사항은 수문모형에 사용되는 강우의 빈도와 유출의 빈도가 동일하다는 가정에 근거한다. 즉, 유역의 초기함수조건, 강우강도, 강우의 시간적 분포와 관계없이 동일한 빈도로 고려되는 문제점이 있다. 이러한 점에서 비교적 장기간의 자료를 확보하고 있는 계측유역에 대해서 다변량 확률밀도함수를 적용하여 비선형관계를 고려한 수문빈도해석기법을 개발하고자 한다. 본 연구에서는 이변량 분석기법(bivariate analysis) 중 전통적인 이변량 분포에 비해 주변분포형(marginal distribution)을 자유롭게 선택할 수 있는 장점이 있는 추계학적 Copula 모형을 활용하여 댐 및 저수지 상류유역의 강우량과 유입량을 대상으로 이변량 분석을 수행하고자 한다. 최종적으로 비선형 관계에 있는 강수량과 유출량 사이에 이변량 빈도해석 모형을 개발하고 기존 해석방법과의 종합적인 비교를 실시하였다.

• Journal of Broadcast Engineering
• /
• v.25 no.2
• /
• pp.157-165
• /
• 2020

In this paper. we propose an effective compensation scheme to the quantization error arisen from quantized learning in a machine learning on an embedded system. In the machine learning based on a gradient descent or nonlinear signal processing, the quantization error generates early vanishing of a gradient and occurs the degradation of learning performance. To compensate such quantization error, we derive an orthogonal compensation vector with respect to a maximum component of the gradient vector. Moreover, instead of the conventional constant learning rate, we propose the adaptive learning rate algorithm without any inner loop to select the step size, based on a nonlinear optimization technique. The simulation results show that the optimization solver based on the proposed quantized method represents sufficient learning performance.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.10 no.6
• /
• pp.45-55
• /
• 1973

This paper discusses the nonlinear time-invarient circuit composed of a tunnel diode. Prior to determine the solution of the nonlinear network which has negative resistance elements, the static characteristics of the nonlinear resistance elements need to be represented by function. Polynomial curve fitting is discussed to represent the static characteristies by least squares approximation. In order to solve the nonlinear network, the state equations for the networks are set up and solved by prediction corrector method. Finally, the limit cycle is plotted to discuss the stability of the nonlinear network and the oscillation condition.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.26 no.3B
• /
• pp.279-289
• /
• 2006

The reconstruction of low dimension nonlinear behavior from the hydrologic time series has been an active area of research in the last decade. In this study, we present the applications of a powerful state space reconstruction methodology using the method of Support Vector Machines (SVM) to the Great Salt Lake (GSL) volume. SVMs are machine learning systems that use a hypothesis space of linear functions in a Kernel induced higher dimensional feature space. SVMs are optimized by minimizing a bound on a generalized error (risk) measure, rather than just the mean square error over a training set. The utility of this SVM regression approach is demonstrated through applications to the short term forecasts of the biweekly GSL volume. The SVM based reconstruction is used to develop time series forecasts for multiple lead times ranging from the period of two weeks to several months. The reliability of the algorithm in learning and forecasting the dynamics is tested using split sample sensitivity analyses, with a particular interest in forecasting extreme states. Unlike previously reported methodologies, SVMs are able to extract the dynamics using only a few past observed data points (Support Vectors, SV) out of the training examples. Considering statistical measures, the prediction model based on SVM demonstrated encouraging and promising results in a short-term prediction. Thus, the SVM method presented in this study suggests a competitive methodology for the forecast of hydrologic time series.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 1999.11a
• /
• pp.214-217
• /
• 1999

In this paper we proposed a heterogeneous hidden layer consisting of both sigmoid functions and RBFs(Radial Basis Function) in multi-layered neural networks. Focusing on the orthogonal relationship between the sigmoid function and its derivative, a derived RBF that is a derivative of the sigmoid function is used as the RBF in the neural network. so the proposed neural network is called ONN's feasibility Neural Network). Identification results using a nonlinear. function confirm both the ONN's feasibility and characteristics by comparing with those obtained using a conventional neural network which has sigmoid function or RBF in hidden layer.

• The Korean Journal of Applied Statistics
• /
• v.3 no.2
• /
• pp.91-105
• /
• 1990

Several practical prior distributions are derived using the maximum entropy principle. Also, an interactive method for estimating a prior distribution which uses the minimum cross-entropy principle is proposed when there are many prior informations. The consistency of the prior distributions obtained by the entropy principles is discussed.

• Proceedings of the Korean Society of Coastal and Ocean Engineers Conference
• /
• 1993.07a
• /
• pp.178-182
• /
• 1993

유속의 연직분포를 알기 위한 방안의 하나로 Galerkin 함수 전개모델에 대한 연구가 활발히 진행되어 왔다(Heaps, 1972； Davies, 1980； Cordon,1982； 최,1983；강등,1993). Galerkin 함수 이용 3차원 모델은 수직좌표상의 수평유속 성분을 수평좌표, 시간에 따라 변하는 계수 와 수심에 따라 변하는 함수군의 곱의 형태로 선형전개하며, 계수의 값은 Galerkin 방법을 써서구한다. (중략)

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.12 no.11
• /
• pp.5164-5171
• /
• 2011

In fuzzy modeling for nonlinear process, typically using the given data, the fuzzy rules are formed by the input variables and the space division by selecting the input variable and dividing the input space for each input variables. The premise part of the fuzzy rule is identified by selection of the input variables, the number of space division and membership functions and the consequent part of the fuzzy rule is identified by polynomial functions in the form of simplified and linear inference. In general, formation of fuzzy rules for nonlinear processes using the given data have the problem that the number of fuzzy rules exponentially increases. To solve this problem complex nonlinear process can be modeled by separately forming the fuzzy rules by means of fuzzy division of each input space. Therefore, this paper utilizes individual input space to generate fuzzy rules. The premise parameters of the fuzzy rules are identified by Min-Max method using the minimum and maximum values of input data set and membership functions are used as a series of triangular, gaussian-like, trapezoid-type membership functions. And lastly, using the data which is widely used in nonlinear process we evaluate the performance and the system characteristics.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.1
• /
• pp.1-10
• /
• 2022

Deep neural networks are widely used to solve various problems. In a fully connected neural network, the nonlinear activation function is a function that nonlinearly transforms the input value and outputs it. The nonlinear activation function plays an important role in solving the nonlinear problem, and various nonlinear activation functions have been studied. In this study, we propose a combined parametric activation function that can improve the performance of a fully connected neural network. Combined parametric activation functions can be created by simply adding parametric activation functions. The parametric activation function is a function that can be optimized in the direction of minimizing the loss function by applying a parameter that converts the scale and location of the activation function according to the input data. By combining the parametric activation functions, more diverse nonlinear intervals can be created, and the parameters of the parametric activation functions can be optimized in the direction of minimizing the loss function. The performance of the combined parametric activation function was tested through the MNIST classification problem and the Fashion MNIST classification problem, and as a result, it was confirmed that it has better performance than the existing nonlinear activation function and parametric activation function.

• Journal of Korea Water Resources Association
• /
• v.39 no.12 s.173
• /
• pp.997-1012
• /
• 2006

We have presented a nonparametric stochastic approach for the SOI(Southern Oscillation Index) series that used nonlinear methodology called Nonlinear AutoRegressive(NAR) based on conditional kernel density function and CAFPE(Corrected Asymptotic Final Prediction Error) lag selection. The fitted linear AR model represents heteroscedasticity, and besides, a BDS(Brock - Dechert - Sheinkman) statistics is rejected. Hence, we applied NAR model to the SOI series. We can identify the lags 1, 2 and 4 are appropriate one, and estimated conditional mean function. There is no autocorrelation of residuals in the Portmanteau Test. However, the null hypothesis of normality and no heteroscedasticity is rejected in the Jarque-Bera Test and ARCH-LM Test, respectively. Moreover, the lag selection for conditional standard deviation function with CAFPE provides lags 3, 8 and 9. As the results of conditional standard deviation analysis, all I.I.D assumptions of the residuals are accepted. Particularly, the BDS statistics is accepted at the 95% and 99% significance level. Finally, we split the SOI set into a sample for estimating themodel and a sample for out-of-sample prediction, that is, we conduct the one-step ahead forecasts for the last 97 values (15%). The NAR model shows a MSEP of 0.5464 that is 7% lower than those of the linear model. Hence, the relevance of the NAR model may be proved in these results, and the nonparametric NAR model is encouraging rather than a linear one to reflect the nonlinearity of SOI series.