인공신경망(artificial neural network)을 활용한 일반계 고등학생의 비문학 독해 평가에서의 인지진단모형 적용 방안

이상빈,김태호,현윤진,최숙기 한국독서학회 2018 독서연구 Vol.0 No.46

This study would examine the validity of the result of an analysis of the evaluation of general high school students’ reading comprehension, utilizing a cognitive diagnosis model through an artificial neural network and implement machine learning of an artificial neural network model. For this purpose, this study first extracted cognitive factors Attributes related to capability or knowledge in reading comprehension, developed Q-Matrix and validated this. This study estimated the mastery profile of each cognitive factor of the reading comprehension of non-literary texts with the result of the evaluation of reading comprehension made by producing a tool for the evaluation of reading comprehension based on the developed Q-Matrix, utilizing a cognitive diagnosis model. Next, by running a machine learning process, putting the result of the actual evaluation Score and the proficiency profile of each cognitive factor of the reading comprehension of non-literary texts into the artificial neural network model, respectively, as the input value and output value, this study would develop an algorithm in which the artificial neural network model itself estimates a proficiency profile by cognitive factor according to the result of the test of reading comprehension. In the artificial neural network model in this study, the result of the evaluation of test data was lower, though slightly than the result of the evaluation of training data, so it turned out that complete machine learning did not take place; however, since prediction accuracy exceeded 95%, and kappa coefficient exceeded 0.9 (Excellent conformity degree), it cannot be considered to be an overfitting result of training data, so it is judged that this study presented the possibility of machine learning. 본 연구는 인지진단모형(Cognitive Diagnosis Model)을 활용한 일반계 고등학생의 독해 평가 분석 결과를 인공신경망을 통하여 그 타당성을 검토하고, 나아가 인공신경망 모형의 머신러닝(Machine Learning)을 구현해보고자 하였다. 이를 위해 먼저, 독해 기능이나 지식과 관련한 인지요소(attribute)를 추출하여 Q-행렬(Q-Matrix)을 개발하고 이를 타당화 하였다. 이렇게 개발된 Q-행렬에 기반하여 독해 평가 도구를 제작하여 시행한 독해 평가 결과를 인지진단모형을 활용하여 비문학 독해 인지요소별 숙달(mastery) 프로파일을 추정하였다. 다음으로 실제 평가 결과(Score)와 비문학 독해 인지요소별 숙달 프로파일을 인공신경망 모형의 각각 입력값과 출력값으로 투입하여 머신러닝 과정을 실행함으로써, 인공신경망 모형이 독해 평가결과에 따라 인지요소별 숙달 프로파일을 스스로 추정하는 알고리즘을 개발하고자 하였다. 본 연구의 인공신경망 모형은 훈련데이터 평가결과에 비해 테스트 데이터 평가 결과가 약간이지만 낮은 값을 보여 완전한 머신러닝이 일어나지는 않은 것으로 나타났지만, 예측 정확도가 95%를 넘었고, 카파계수가 0.9(훌륭한 일치도)를 넘었기 때문에 훈련 데이터에 대한 과적합의 결과로 볼 수 없으므로, 머신러닝의 가능성을 제시하였다고 할 수 있다.

MLP, GR, RBF 인공신경망을 이용한 분자 동역학 데이터 기반 초탄성 구성방정식 모델링

김수한,왕호림,이원주,안병혁,김유정,이주호,신현성 대한기계학회 2023 大韓機械學會論文集A Vol.47 No.1

In this study, three neural network models are used to establish neural network constitutive models (NNCMs) for hyperelastic materials: multilayer perceptron neural network, generalized regression neural network, and radial basis function neural network (RBFNN). We focused on artificial neural network training of the dataset: full atomistic molecular dynamics simulation results. We obtained the optimal hyperparameters empirically for each neural network model to construct an optimal NNCM. The three models above were compared based on statistical error analysis. The results indicated that the RBFNN model had the highest accuracy in terms of the root-mean-square error, mean absolute error, and regression values. We applied the proposed NNCM model to data-driven computational mechanics to verify the proposed NNCM model, with the data-driven solver demonstrating superior performance of RBFNN-based NNCM over those of the other neural network models. 본 연구에서는 다층 퍼셉트론 신경망(MLPNN), 일반화 회귀 신경(GRNN) 및 방사형 기저 함수 신경망(RBFNN)의 세 가지 다른 신경망 모델을 사용하여 초탄성 물질에 대한 신경망 구성방정식 모델(NNCM)을 구축하였다. 분자 동역학(MD) 데이터셋을 학습에 활용하였으며, 최적의 NNCM의 구축을 위해 경험적으로 최적의 하이퍼파라미터를 선정하였다. 위의 세 가지 모델을 통계적 오류 분석을 기반으로 비교하였다. 그 결과 RBFNN 모델이 RMSE, MAE 및 R 값 측면에서 다른 모델에 비해 더 적합하다는 것을 확인하였다. 제안된 NNCM 모델을 데이터 기반 전산역학(DDCM)에 적용하여 제안된 NNCM 모델의 성능을 비교하였다.

단계적 회귀분석과 인공신경망 모형을 이용한 광양항 석탄·철광석 물동량 예측력 비교 분석

조상호,남형식,류기진,류동근 한국항해항만학회 2020 한국항해항만학회지 Vol.44 No.3

It is very important to forecast freight volume accurately to establish major port policies and future operation plans. Thus, related studies are being conducted because of this importance. In this paper, stepwise regression analysis and artificial neural network model were analyzed to compare the predictive power of each model on Gwangyang Port, the largest domestic port for coal and iron ore transportation. Data of a total of 121 months January 2009-January 2019 were used. Factors affecting coal and iron ore trade volume were selected and classified into supply-related factors and market/economy-related factors. In the stepwise regression analysis, the tonnage of ships entering the port, coal price, and dollar exchange rate were selected as the final variables in case of the Gwangyang Port coal volume forecasting model. In the iron ore volume forecasting model, the tonnage of ships entering the port and the price of iron ore were selected as the final variables. In the analysis using the artificial neural network model, trial-and-error method that various Hyper-parameters affecting the performance of the model were selected to identify the most optimal model used. The analysis results showed that the artificial neural network model had better predictive performance than the stepwise regression analysis. The model which showed the most excellent performance was the Gwangyang Port Coal Volume Forecasting Artificial Neural Network Model. In comparing forecasted values by various predictive models and actually measured values, the artificial neural network model showed closer values to the actual highest point and the lowest point than the stepwise regression analysis. 항만의 주요 정책 및 향후 운영계획 수립 시 정확한 물동량 예측에 관한 연구는 매우 중요하며 이러한 중요성으로 인해 관련 연구가 활발히 수행되고 있다. 본 논문에서는 국내 최대 석탄 및 철광석 처리 항만인 광양항을 대상으로 단계적 회귀분석과 인공신경망모형을 활용하여 모형간 예측력을 비교하였다. 2009년 1월부터 2019년 1월까지 총 121개월의 월별자료를 활용하였으며 석탄 및 철광석 물동량에 영향을 주는 요인을 선정하여 공급관련요인과 시장·경제관련요인으로 분류하였다. 단계적 회귀분석 결과, 광양항 석탄 물동량 예측모형의 경우, 입항선박 톤수, 석탄가격 및 대미환율이 최종변수로 선정되었고 철광석 물동량 예측모형의 경우, 입항선박 톤수, 철광석가격이 최종변수로 선정되었다. 인공신경망모형의 경우, 모델 성능에 영향을 미치는 다양한 Hyper-parameters를 조정하며 최적 모델을 선정하는 시행착오법을 사용하였다. 분석결과 인공신경망모형이 단계적 회귀분석에 비해 우수한 예측성능을 나타내었으며 예측 모형별 예측값과 실측값을 그래프 상 비교 시에도 인공신경망모형이 단계적 회귀분석에 비해 고·저점을 유사하게 나타냈다.

Improving the Performance of Industrial Boiler Using Artificial Neural Network Modeling and Advanced Combustion Control

Yul Yunazwin Nazaruddin,Abdullah Nur Aziz,Wisnu Sudibjo 제어로봇시스템학회 2008 제어로봇시스템학회 국제학술대회 논문집 Vol.2008 No.10

In heat generation process, performance improvement is a critical factor and essential. An alternative solution is by designing an advanced combustion controller based on neural-predictive control strategy. However, for accomplishing such goal it requires adequate boiler model as well as combustion model. Although heat transfer and combustion processes in boiler are too complex to be analytically described with mathematical model, it can be approximated by artificial neural network model. This paper presents an alternative strategy to model the boiler and combustion process as well as proposes an advanced control strategy that takes the advantage of artificial neural network’ ability as a universal function approximation. A feedforward neural network algorithm is applied to construct the models and the gradient descent technique seeks the optimal network weights, from which the nonlinear predictive control law under the reduced excess air level is derived. Direct application of this control strategy to real-time data taken from a running boiler system at an oil refinery plant demonstrated the benefit of the algorithm to improve the boiler combustion performance.

중소하천유역에서 유출량 산정을 위한 NEURAL NETWORK의 적용

최윤영 ( Yun Young Choi ) 한국수처리학회 2003 한국수처리학회지 Vol.11 No.1

N/A In order to resolve the rainfall-runoff forecast model s uncertainty of model parameters and to increase the model s output, the study utilized Neural Networks model such as ANN and GRNN model. which mathematically interpret human thought processes. In order to calibrate and verify the runoff forecast model, seven flood events(1997-2002) observed at the Kumho water level gage station located on the midstream of Kumho river were chosen, of which five were used as calibration data and two as that for verification. First, as a result of applying ANN model, of the ten models, the ANN-8-8(10 input layer nodes, 8 nodes in the first hidden layer, 8 nodes in the second hidden layer, 1 output layer node) model was found to be the more suitable model in an actual hydro-event. Second, as a result of applying GRNN model, of the nine models, the GN-4a(l0 input layer nodes, 4 nodes in the pattern layer, spread σ=0.05, 1 output layer node) model was considered suitable. In addition, the numbers of pattern layer node increased, but it didn`t increase the learning rate. Finally, according to the results of the statistical analysis of the ANN and GRNN model, it was shown that ANN model was better. However, since there are numerous difficulties in determining the superiority of a model based on data only from the seven heavy rainfall events used in this study, it is judged that continued application and study based on more quantitative and qualitative data are necessary.

‘인공 지능이 인간답다’라 함은? -심층 신경망 언어 모델을 중심으로-

구건우,이재민,김유영,임선희,전수경,최릉운,박명관 동국대학교 동서사상연구소 2022 철학·사상·문화 Vol.- No.39

The statement 'artificial intelligence is human-like' has recently been a buzzword as an AI/neural-network language model is reported to reach a human level in language learning and processing, with the model implemented by applying the so-called deep neural network to language learning. In this paper, we evaluate the proposition of 'artificial intelligence is human', concentrating on the neural network language model. Accordingly, when neural network language models exhibit aspects of language learning and processing that humans do, we need to clarify whether the two human and AI learning & processing systems are homologous or what differences there are between them. Furthermore, based on the comparison of the two systems, we discuss what lights neural network language models shed on traditional issues raised for human language acquisition and processing, such as nature vs. nurture, human linguistic ability, etc. 사람의 신경망을 컴퓨터 공학 기술로 구현한 것이 소위 심층 신경망이다. 이를 적용한 언어 모델이 사람의 언어 학습을 재현함에 따라, 인공 지능 신경망 언어 모델이 언어 학습과 처리에 있어 사람에 상응하는 수준에 도달하고 있다고 보고되면서 ‘인공 지능(신경망 언어 모델)이 (언어 학습과 처리에 있어) 인간답다’라는 말이 회자된다. 본고에서는 신경망 언어 모델을 중심으로 ‘인공 지능이 인간답다’라는 명제를 재검토한다. 이에 따라, 신경망 언어 모델이 사람과 같은 언어 학습과 처리의 양상을 보인다고 할 때, 두 학습 및 처리 시스템이 상동한지 아니면 어떤 차이점이 있는지를 해명한다. 나아가서, 두 시스템의 비교를 바탕으로, 사람을 대상으로 언어 학습과 처리에 관한 전통적 이슈들에 관하여, 사람과 대비하여 신경망 언어 모델은 어떤 특성을 보이는지 논의한다.

신경회로망을 이용한 비선형모델기준 적응제어기의 설계에 관한 연구

金昌圭,卓漢浩 진주산업대학교 1997 산업과학기술연구소보 Vol.- No.4

Artificial neural networks show promise as elements in control systems. One feature of traditional control system design which has been largely lacking in the work with neural networks to date is the analysis of the closed loop stability of the controlled system. The principal aim of this work was to develop a control architecture for which such analyses can be made Theoretical results were developed for the stability of systems using controllers based on approximate controllers, such as artificial neural networks. These results are based on fairly strong assumptions concerning the ability of the artificial neural networks to learn a model of the forward dynamics of the plant. The stability result are based on an application of the concepts of Liapunov stability to systems which are stable in the sense of Lagrange: that is, the stability is with respect to a region in the state space, rather than a point. These stability results motivate the selection of a model reference adaptive controller architecture. The controller incorporates a performance model, which provides a desired output trajectory in response to system inputs (commands), and a convergence model, which determines the desired perturbation dynamics of the true output about the model output trajectory. The conotrol input is selected by on line minimization of a cost function, based on a Liapunov-like function derived from the convergence model. This minimization procedure uses the neural networks model of the system dynamics to predict the response of the system to a candidate control input. The controller described was applied to the adaptive control of the inverted pendulum problem. The addition of a dither signal to the calculated control input was shown to enhance the ability of error back propagation to improve the artificial neural network model on line.

Performance of an Artificial Neural Network Model for Simulating Saltwater Intrusion Process in Coastal Aquifers when Training with Noisy Data

Rajib Kumar Bhattacharjya,Bithin Datta,Mysore G. Satish 대한토목학회 2009 KSCE JOURNAL OF CIVIL ENGINEERING Vol.13 No.3

This paper evaluates the performance of an Artificial Neural Networks (ANN) model for approximating density depended saltwater intrusion process in coastal aquifer when the ANN model is trained with noisy training data. The data required for training, testing and validation of the ANN model are generated using a numerical simulation model. The simulated data, consisting of corresponding sets of input and output patterns are used for training a multilayer perception using back-propagation algorithm. The trained ANN predicts the concentration at specified observation locations at different time steps. The performance of the ANN model is evaluated using an illustrative study area. These evaluation results show the efficient predicting capabilities of an ANN model when trained with noisy data. A comparative study is also carried out for finding the better transfer function of the artificial neuron and better training algorithms available in Matlab for training the ANN model. This paper evaluates the performance of an Artificial Neural Networks (ANN) model for approximating density depended saltwater intrusion process in coastal aquifer when the ANN model is trained with noisy training data. The data required for training, testing and validation of the ANN model are generated using a numerical simulation model. The simulated data, consisting of corresponding sets of input and output patterns are used for training a multilayer perception using back-propagation algorithm. The trained ANN predicts the concentration at specified observation locations at different time steps. The performance of the ANN model is evaluated using an illustrative study area. These evaluation results show the efficient predicting capabilities of an ANN model when trained with noisy data. A comparative study is also carried out for finding the better transfer function of the artificial neuron and better training algorithms available in Matlab for training the ANN model.

Real-time forecasting of wave heights using EOF – wavelet – neural network hybrid model

Oh, Jihee,Suh, Kyung-Duck Elsevier 2018 Ocean engineering Vol.150 No.-

<P><B>Abstract</B></P> <P>Recently, along with the development of data-driven models, artificial neural networks (ANN) have been used in ocean wave forecasting models. Hybridization of ANN with wavelet analysis or fuzzy logic approach has also been used. The wavelet and neural network hybrid models (WNN models) show better performance than ANN models. However, their accuracy decreases with increasing lead time because they do not consider the relation between wave and meteorological variables. Moreover, the WNN model has been developed to forecast the wave height at a single location where the past wave height data are available. To resolve these problems, in this paper, a hybrid model is developed by combining the empirical orthogonal function analysis and wavelet analysis with the neural network (abbreviated as EOFWNN model). The past wave height data at multiple locations and the past and future meteorological data in the surrounding area including the wave stations are used as input data. The model then forecasts the wave heights at the locations for various lead times. The developed model is employed to forecast the wave heights at eight wave observation stations in the coastal waters around the East/Japan Sea. The EOFWNN model is shown to perform better compared with the WNN model for all lead times regardless of the decomposition level of wavelet analysis. The EOFWNN model is proven to be a promising tool for forecasting wave heights at multiple locations where the past wave height data and the past and future meteorological data in the surrounding area are available.</P> <P><B>Highlights</B></P> <P> <UL> <LI> A real time wave height forecasting model at multiple locations is developed. </LI> <LI> A hybrid model is developed by combining EOF and wavelet analyses with neural network. </LI> <LI> The model forecasts significant wave heights at multiple locations for various lead times. </LI> <LI> The model is applied to eight wave stations in the coastal waters around the East/Japan Sea. </LI> <LI> The model is compared with a single point wave forecast model. </LI> </UL> </P>

An Application of BP Neural Network to the Prediction of Compressive Strength in Circular Concrete Columns Confined with CFRP

Khalil AL-Bukhaiti,Liu Yanhui,Zhao Shichun,Hussein Abas 대한토목학회 2023 KSCE Journal of Civil Engineering Vol.27 No.7

The neural network comprises many neurons with extensive interconnections operatingparallel and performing specific functions. This paper establishes a BP neural networkprediction model for the compressive strength of CFRP-confined concrete based on a largenumber of experimental data to study the predictive ability of the BP neural network on thecompressive strength of CFRP-confined concrete and the output performance of the neuralnetwork model. The model is based on a BP neural network that has been trained using manyexperimental data. An investigation is being conducted on the effect of different datacombinations on the accuracy of the predictions made by the neural network model. Thehigh-precision BP network model is created into generic and simplified formulae for applicationconvenience. These formulas are developed based on the theory of neural networks. Theneural network models' findings and the empirical formulae for making predictions arecompared and discussed. The BP neural network accurately predicts the compressive strengthof CFRP-confined concrete, with over 90% of its data points having less than 15% error. Incomparison, the regression model shows less accuracy, with less than 70% of its data pointshaving an error within 15%. Compared to traditional regression models, the simple linearequation derived using Purelin instead of Sigmoid as the transfer function only adds a constantterm. The average value of prediction/test results is 1.011. The analysis results show that BPneural network can extract the input and output parameters' data information well and obtaina high-accuracy prediction model. The coefficient of variation is 0.112, which indicates thatthe prediction accuracy and stability are greater than average.