• Annual Conference on Human and Language Technology
• /
• 2017.10a
• /
• pp.111-115
• /
• 2017

목적 지향적 대화 시스템(Goal-oriented dialogue system) 은 텍스트나 음성을 통해 특정한 목적을 수행 할 수 있는 시스템이다. 최근 RNN(recurrent neural networks)을 기반으로 대화 데이터를 end-to-end learning 방식으로 학습하여 대화 시스템을 구축하는데에 활용한 연구가 있다. End-to-end 방식의 학습은 도메인에 대한 지식 없이 학습 데이터 자체만으로 대화 시스템 구축을 위한 학습이 가능하다는 장점이 있지만 도메인 지식을 학습하기 위해서는 많은 양의 데이터가 필요하다는 단점이 존재한다. 이에 본 논문에서는 도메인 특정 지식을 결합하여 end-to-end learning 방식의 학습이 가능한 Hybrid Code Network 구조를 기반으로 한국어로 구성된 식당 예약에 관련한 대화 데이터셋을 이용하여 식당 예약을 목적으로하는 대화 시스템을 구축하는 방법을 제안한다. 실험 결과 본 시스템은 응답 별 정확도 95%와 대화 별 정확도 63%의 성능을 나타냈다.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.09a
• /
• pp.307-310
• /
• 2003

In the maritime container terminal, LMTT(Linear Motor-based Transfer Technology) is horizontal transfer system for the yard automation, which has been proposed to take the place of AGV(Automated Guided Vehicle). The system is based on PMLSM (Permanent Magnetic Linear Synchronous Motor) that is consists of stator modules on the rail and shuttle car (mover). Because of large variant of mover's weight by loading and unloading containers, the difference of each characteristic of stator modules, and a stator module's trouble etc., LMCPS (Linear Motor Conveyance Positioning System) is considered as that the system is changed its model suddenly and variously. In this paper, we will introduce the soft-computing method of a multi-step prediction control for LMCPS using DR-FNN (Dynamically-constructed Recurrent Fuzzy Neural Network). The proposed control system is used two networks for multi-step prediction. Consequently, the system has an ability to adapt for external disturbance, cogging force, force ripple, and sudden changes of itself.

• Asia pacific journal of information systems
• /
• v.7 no.1
• /
• pp.67-83
• /
• 1997

The utility of learning techniques in investment analysis has been demonstrated in many areas, ranging from forecasting individual stocks to entire market indexes. To date, however, the application of artificial intelligence to financial forecasting has focused largely on short predictive horizons. Usually the forecast window is a single period ahead; if the input data involve daily observations, the forecast is for one day ahead; if monthly observations, then a month ahead; and so on. Thus far little work has been conducted on the efficacy of long-term prediction involving multiperiod forecasting. This paper examines the impact of alternative procedures for extended prediction using knowledge discovery techniques. One dimension in the study involves temporal granularity: a single jump from the present period to the end of the forecast window versus a web of short-term forecasts involving a sequence of single-period predictions. Another parameter relates to the numerosity of input variables: a technical approach involving only lagged observations of the target variable versus a fundamental approach involving multiple variables. The dual possibilities along each of the granularity and numerosity dimensions entail a total of 4 models. These models are first evaluated using neural networks, then compared against a multi-input jump model using case based reasoning. The computational models are examined in the context of forecasting the S&P 500 index.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2006.05a
• /
• pp.301-306
• /
• 2006

본 연구에서는 강우의 시 공간적 분포의 불규칙한 변동성을 고려한 강우-유출예측을 위해 인공신경망(Artificial Neural Networks: ANNs)의 기법의 일종인 자기조직화(Self Organizing Map: SOM) 이론과 역전파 학습 알고리즘(Back Propagation Algorithm: BPA) 이론을 복합적으로 이용하였다. 기존의 인공신경망 연구에서 야기된 저..갈수기의 유출량에 대한 과대평가, 홍수기의 유출량에 대한 과소평가, 예측값이 선행 유출량의 지속성을 갖는 Persistence 현상을 해결하기 위하여 패턴분류 성능을 지닌 SOM 이론을 도입하여 예측모형의 전처리 과정으로 이용하였다. 이는 기존의 인공신경망 모형이 하나의 모형을 구성하여 유출량의 전 범위에 해당하는 자료를 예측하는 방법을 개선한 것으로 SOM에 의해 패턴이 분류된 강우-유출관계의 각 패턴별 예측모형을 통해 분류된 자료들의 예측을 수행하는 방법이다. 이와 같이 SOM을 강우-유출예측모형의 전처리과정으로 이용함으로서 기존의 인공신경망 연구에서 야기된 현상들을 해결할 수 있었고, 예측력 또한 기존의 인공신경망 모형의 결과에 비해 우수하였다.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2009.05a
• /
• pp.314-318
• /
• 2009

본 연구의 목적은 강우-유출자료 및 댐 방류량 자료의 비선형적인 특정을 가장 잘 반영할 수 있는 신경망모형을 적용하여 수질정책의 기초자료를 제공하기 위하여 신뢰성 있는 유량자료를 산정하는 모형을 개발하였고 이를 낙동강 유역에 적용하는 것이다. 이를 위해서 낙동강물환경연구소의 8일 측정 유량이 가지는 정확성을 이용하면서 상류 댐의 일 방류량자료와 유역별 강우자료 및 국토해양부 수위관측소의 수위자료를 연계하여 유량을 보간할 수 있는 유량 보간 신경망 모형을 개발하였다. 신경망 모형의 출력값은 낙동강물환경 연구소에서 측정하지 않은 기간에 대하여 유량을 보간할 수 있도록 구성하였으며 신경망 모형의 구조는 입력층과 출력층 사이에 하나의 은닉층이 존재하는 다층 신경망으로 구성하였으며, 학습단계에서는 오류 역전파 알고리듬 학습방법 중 모멘텀법을 사용하였다. 본 연구를 통하여 낙동강 전 유역에 대하여 유량 보간 모형을 적용한 결과 댐 방류량과 강우자료 및 상류 수위 관측소의 유량 자료를 이용한 유량 보간 신경망모형의 일 유량결과의 적용가능성을 검증할 수 있으며, 제시된 모형은 지속적인 수문자료의 질적 향상과 유출패턴의 축적으로 그 성능을 향상시킬 수 있을 것이며 또한 홍수기의 더 정확한 유량예측을 위한 적용사례의 확장 및 SWAT을 이용한 모형의 적용에 대한 연구가 병행되어야 할 것이다.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 1996.10a
• /
• pp.199-202
• /
• 1996

Although numerous studies demonstrate that one technique outperforms the others for a given data set, there is often no way to tell a priori which of these techniques will be most effective to solve a specific problem. Alternatively, it has been suggested that a better approach to classification problem might be to integrate several different forecasting techniques by combining their results. The issues of interest are how to integrate different modeling techniques to increase the prediction performance. This paper proposes the post-model integration method, which means integration is performed after individual techniques produce their own outputs, by finding the best combination of the results of each method. To get the optimal or near optimal combination of different prediction techniques. Genetic Algorithms (GAs) are applied, which are particularly suitable for multi-parameter optimization problems with an objective function subject to numerous hard and soft constraints. This study applied three individual classification techniques (Discriminant analysis, Logit and Neural Networks) as base models to the corporate failure prediction context. Results of composite prediction were compared to the individual models. Preliminary results suggests that the use of integrated methods will offer improved performance in business classification problems.

• Geomechanics and Engineering
• /
• v.13 no.2
• /
• pp.237-254
• /
• 2017

Estimation of slope stability is a very important task in geotechnical engineering. However, its estimation using conventional and soft computing methods has several drawbacks. Use of conventional limit equilibrium methods for the evaluation of slope stability is very tedious and time consuming, while the use of soft computing approaches like Artificial Neural Networks and Fuzzy Logic are black box approaches. Multiple Regression (MR) analysis provides an alternative to conventional and soft computing methods, for the evaluation of slope stability. MR models provide a simplified equation, which can be used to calculate critical factor of safety of slopes without adopting any iterative procedure, thereby reducing the time and complexity involved in the evaluation of slope stability. In the present study, a multiple regression model has been developed and tested its accuracy in the estimation of slope stability using real field data. Here, two separate multiple regression models have been developed for dry and wet slopes. Further, the accuracy of these developed models have been compared and validated with respect to conventional limit equilibrium methods in terms of Mean Square Error (MSE) & Coefficient of determination ($R^2$). As the developed MR models here are not based on any region specific data and covers wide range of parametric variations, they can be directly applied to any real slopes.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.24 no.5 s.176
• /
• pp.1343-1351
• /
• 2000

Thermally induced errors have been significant factors affecting the machine tool accuracy. In this paper, the spindle thermal error has been focused, where the 5 degree of freedom thermal error components are considered. An effective measurement system has been devised for the 5 DOF thermal errors, consisting of gap sensors and thermocouples around the micro-computer interfaced environment. Several thermal error modeling techniques are also implemented for the thermal error prediction: multiple linear regression, neural network and system identification methods, etc. The performance of the thermal error modeling techniques is evaluated and compared, giving the system identification method as the optimum model having the least deviation. The developed system for the thermal error measurement and modeling was practically applied to a CNC machining center, and the spindle thermal errors were effectively compensated around the micro computer-machine tool interfaced networks. The machine tool accuracy was improved about 4-5 times typically.

• Journal of Aerospace System Engineering
• /
• v.7 no.1
• /
• pp.8-18
• /
• 2013

Recently, the advanced condition monitoring methods such as the model-based method and the artificial intelligent method have been applied to maximize the availability as well as to minimize the maintenance cost of the aircraft gas turbines. Among them the non-linear GPA(Gas Path Analysis) method and the GA(Genetic Algorithms) have lots of advantages to diagnose the engines compared to other advanced condition monitoring methods such as the linear GPA, fuzzy logic and neural networks. Therefore this work applies both the non-linear GPA and the GA to diagnose AE3007 turbofan engine for an aircraft, and in case of having sensor noise and bias it is confirmed that the GA is better than the GPA through the comparison of two methods.

• Smart Media Journal
• /
• v.9 no.3
• /
• pp.80-89
• /
• 2020

Out - of - distribution (OOD) samples are frequently encountered when deploying a classification model in plenty of real-world machine learning-based applications. Those samples are normally sampling far away from the training distribution, but many classifiers still assign them high reliability to belong to one of the training categories. In this study, we address the problem of removing OOD examples by estimating marginal density estimation using variational autoencoder (VAE). We also investigate other proper methods, such as temperature scaling, Gaussian discrimination analysis, and label smoothing. We use Chonnam National University (CNU) weeds dataset as the in - distribution dataset and CIFAR-10, CalTeach as the OOD datasets. Quantitative results show that the proposed framework can reject the OOD test samples with a suitable threshold.