• Annual Conference on Human and Language Technology
• /
• 1989.10a
• /
• pp.101-106
• /
• 1989

음성 및 문자를 통한 컴퓨터와의 정보 교환을 위한 통합 사용자 인터페이스 (Intelligent Man- Machine interface) 시스템의 일환으로 한국어 단모음의 인식을 위한 시스템을 인공 신경망 모델을 사용하여 구현하였으며 인식시스템의 상위 접속부에 필요한 단어 인식 모듈에 있어서의 인지 실험도 행하였다. 모음인식의 입력으로는 제1, 제2, 제3 포르만트가 사용되었으며 실험대상은 한국어의 [아, 어, 오, 우, 으, 이, 애, 에]의 8 개의 단모음으로 하였다. 사용한 인공 신경망 모델은 Multilayer Perceptron 이며, 학습 규칙은 Generalized Delta Rule 이다. 1 인의 남성 화자에 대하여 약 94%의 인식율을 나타내었다. 그리고 음성 인식시의 인지 현상 실험을 위하여 약 20개의 단어를 인공신경망의 어휘레벨에 저장하여 음성의 왜곡, 인지시의 lexical 영향, categorical percetion등을 실험하였다. 이때의 인공 신경망 모델은 Interactive Activation and Competition Model을 사용하였으며, 음성 입력으로는 가상의 음성 피쳐 데이타를 사용하였다.

• Journal of the Korea Computer Industry Society
• /
• v.8 no.4
• /
• pp.281-288
• /
• 2007

In this paper, we set 6 interesting points around lips. Analyzed and characterized is the distance change of these 6 interesting points when people pronounces 5 vowels of Korean language. 450 data are gathered and analyzed. Based on this analysis, the system is constructed and the recognition experiments are performed. In this system, we used the camera connected to computer to measure the distance vector between 6 interesting points. In the experiment, 80 normal persons were sampled. The observational error between samples was corrected using normalization method. We analyzed with 30 persons and experimented with 50 persons. We constructed three recognition systems and of those the neural net system gave the best recognition result of 87.44 %.

• Journal of Multimedia Information System
• /
• v.6 no.4
• /
• pp.225-234
• /
• 2019

Deep learning has become increasingly popular in both academic and industrial areas nowadays. Various domains including pattern recognition, Computer vision have witnessed the great power of deep neural networks. However, current studies on deep learning mainly focus on quality data sets with balanced class labels, while training on bad and imbalanced data set have been providing great challenges for classification tasks. We propose in this paper a method of data analysis-based data reduction techniques for selecting good and diversity data samples from a large dataset for a deep learning model. Furthermore, data sampling techniques could be applied to decrease the large size of raw data by retrieving its useful knowledge as representatives. Therefore, instead of dealing with large size of raw data, we can use some data reduction techniques to sample data without losing important information. We group PCB characters in classes and train deep learning on the ResNet56 v2 and SENet model in order to improve the classification performance of optical character recognition (OCR) character classifier.

• Nuclear Engineering and Technology
• /
• v.52 no.6
• /
• pp.1188-1200
• /
• 2020

In this paper, we consider a SDMS (Self-Diagnostic Monitoring System) for a reciprocating pump for the purpose of not only diagnosis but also prognosis. We have replaced a multi class estimator that selects only the most probable one with a multi label estimator such that we are able to see the state of each of the components. We have introduced a measure called certainty so that we are able to represent the symptom and its state. We have built a flow loop for a reciprocating pump system and presented some results. With these changes, we are not only able to detect both the dominant symptom as well as others but also to monitor how the degree of severity of each component changes. About the dominant ones, we found that the overall recognition rate of our algorithm is about 99.7% which is slightly better than that of the former SDMS. Also, we are able to see the trend and to make a base to find prognostics to estimate the remaining useful life. With this we hope that we have gone one step closer to the final goal of prognosis of SDMS.

• Proceedings of the Korean Society for Cognitive Science Conference
• /
• 2000.05a
• /
• pp.311-321
• /
• 2000

본 논문에서는 정신분석과 두 종류의 정서이론, 인공지능과 신경회로망 그리고 퍼지 페트리 네트 등을 사용하여 사람의 인지과정을 모방한 인지모형시스템을 개발하였다. 먼저 프로이트의 정신분석을 사용하여 정신의 구조를 그래프로 표현한 후 이것을 '마음의 지도'라 명명하였다. 인지모형시스템을 구현하기 위한 첫 번째 작업으로 동적인 추론을 할 수 있는 지능 모델인 KNBN(Kohonen Network based Belief Network)을 제안하였다. KNBN으로 표현한 마음의 약도 내에서 연결강도 값으로 사용할 상대적 데이터를 만들기 위한 근거로서는 '정서'를 사용하였는데, 플라칙의 진화론에 근거한 정서이론과 오토니의 인지적 정서이론을 결합하여 데이터로 만든후 이 수치를 연결강도로 사용하였다. 이 두 개의 정서이론을 결합하는 알고리즘을 만들기 위해 페트리네트를 변형한 퍼지 페트리네트를 제안하였다. 또한 오토니가 주장하는 정서의 인지구조를 사람들이 그대로 이해하는지 여부를 알기 위해 대학생 100명을 대상으로 설문지를 사용해 정서의 인지구조에 대해 조사하였고 그 결과 값에 근거하여 두 개의 정서이론 결합 알고리즘을 만들었다. 이것으로 정서 발화에 대한 상대적인 수치가 산출되었고, 이것을 KNBN으로 표현한 마음의 약도에 결합하기 위해 0과 1사이의 수치로 정규화 하였다. 이렇게 정규화된 데이터를 이용해 인지 모형 시스템을 개발하였다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.16 no.5
• /
• pp.403-412
• /
• 1991

This paper proposes a type of self organizing neural network which recognizes arbitrary symbols as well as numerical or alphabetic characters. The proposed algorithm autonomically organizes and classifies similar patterns on the basis of the distribution types of characteristics in the input images. Thus it can be appliced for the recognition of arbitrary images when it is difficult to establish a learning rule. It performs a stale recognition process with in the limit of the memory capacity. The cheme was applied and tested to 50 different image patterns with increased noise level up to 44%(SNR 2dB). The implementation results demonstrate that the proposed algorithm successfully recognizes the image patterns changed due to the various noise levels and thus proves excellent antinoise characteristics.

• Journal of Korean Institute of Industrial Engineers
• /
• v.26 no.3
• /
• pp.193-199
• /
• 2000

In many studies, considerable attention has been focussed upon choosing a model which represents underlying process of time series and forecasting the future. In the real world, however, there may be some cases that one model can not reflect all the characteristics of original time series. Under such circumstances, we may get better performance by combining the forecasts from several models. The most popular methods for combining forecasts involve taking a weighted average of multiple forecasts. But the weights are usually unstable. In cases the assumptions of normality and unbiasedness for forecast errors are satisfied, a Bayesian method can be used for updating the weights. In the real world, however, there are many circumstances the Bayesian method is not appropriate. This paper proposes a PNN(Probabilistic Neural Net) approach as a method for combining forecasts that can be applied when the assumption of normality or unbiasedness for forecast errors is not satisfied. In this paper, PNN method, which is similar to Bayesian approach, is suggested as an updating method of the unstable weights in the combination of the forecasts. The PNN method has been usually used in the field of pattern recognition. Unlike the Bayesian approach, it requires no assumption of a specific prior distribution because it gets probabilities by using the distribution estimated from given data. Empirical results reveal that the PNN method offers superior predictive capabilities.

• Nuclear Engineering and Technology
• /
• v.46 no.1
• /
• pp.101-108
• /
• 2014

About 40% of reactors in the world are being operated beyond design life or are approaching the end of their life cycle. During long-term operation, various degradation mechanisms occur. Fatigue caused by alternating operational stresses in terms of temperature or pressure change is an important damage mechanism in continued operation of nuclear power plants. To monitor the fatigue damage of components, Fatigue Monitoring System (FMS) has been installed. Most FMSs have used Green's Function Approach (GFA) to calculate the thermal stresses rapidly. However, if temperature-dependent material properties are used in a detailed FEM, there is a maximum peak stress discrepancy between a conventional GFA and a detailed FEM because constant material properties are used in a conventional method. Therefore, if a conventional method is used in the fatigue evaluation, thermal stresses for various operating cycles may be calculated incorrectly and it may lead to an unreliable estimation. So, in this paper, the modified GFA which can consider temperature-dependent material properties is proposed by using an artificial neural network and weight factor. To verify the proposed method, thermal stresses by the new method are compared with those by FEM. Finally, pros and cons of the new method as well as technical findings from the assessment are discussed.

• Nuclear Engineering and Technology
• /
• v.53 no.11
• /
• pp.3563-3579
• /
• 2021

Measured decay heat data of light water reactor (LWR) spent nuclear fuel (SNF) assemblies are adopted to train machine learning (ML) models. The measured data is available for fuel assemblies irradiated in commercial reactors operated in the United States and Sweden. The data comes from calorimetric measurements of discharged pressurized water reactor (PWR) and boiling water reactor (BWR) fuel assemblies. 91 and 171 measurements of PWR and BWR assembly decay heat data are used, respectively. Due to the small size of the measurement dataset, we propose: (i) to use the method of multiple runs (ii) to generate and use synthetic data, as large dataset which has similar statistical characteristics as the original dataset. Three ML models are developed based on Gaussian process (GP), support vector machines (SVM) and neural networks (NN), with four inputs including the fuel assembly averaged enrichment, assembly averaged burnup, initial heavy metal mass, and cooling time after discharge. The outcomes of this work are (i) development of ML models which predict LWR fuel assembly decay heat from the four inputs (ii) generation and application of synthetic data which improves the performance of the ML models (iii) uncertainty analysis of the ML models and their predictions.

• Nuclear Engineering and Technology
• /
• v.53 no.4
• /
• pp.1181-1188
• /
• 2021

When abnormal events occur in a nuclear power plant, operators must conduct appropriate abnormal operating procedures. It is burdensome though for operators to choose the appropriate procedure considering the numerous main plant parameters and hundreds of alarms that should be judged in a short time. Recently, various research has applied deep-learning algorithms to support this problem by classifying each abnormal condition with high accuracy. Most of these models are trained with simulator data because of a lack of plant data for abnormal states, and as such, developed models may not have tolerance for plant data in actual situations. In this study, two approaches are investigated for a deep-learning model trained with simulator data to overcome the performance degradation caused by noise in actual plant data. First, a preprocessing method using several filters was employed to smooth the test data noise, and second, a data augmentation method was applied to increase the acceptability of the untrained data. Results of this study confirm that the combination of these two approaches can enable high model performance even in the presence of noisy data as in real plants.