• Proceedings of the Korea Water Resources Association Conference
• /
• 2018.05a
• /
• pp.424-424
• /
• 2018

인공신경망(Artificial Neural Network; ANN)은 뇌에 존재하는 생물학적 신경세포와 이들의 신호처리 과정을 수학적으로 묘사하여 뇌가 나타내는 지능적 형태의 반응을 구현한 것이다. 인공신경망은 학습(training)을 통해 입력과 출력으로 구성되는 하나의 시스템을 병렬적이고 비선형적으로 구축할 수 있으며, 유연한 모델링 특성으로 인하여 시스템 예측, 패턴인식, 분류 및 공정제어 등의 다양한 분야에서 활용되고 있다. 인공신경망에 대한 최초의 이론은 Muculloch and Pitts(1943)가 제안한 Perceptron에서 시작 되었으며, 기본적인 학습기법인 오차역전파 기법(back-propagation Algorithm) 이 1980년대에 들어 수학적으로 정립된 이후 여러 분야에서 활용되기 시작하였다). 본 연구에서는 하도추적, 구체적으로는 상류단의 복수의 수위관측을 이용하여 하류단의 수위를 예측하기 위하여 인공신경망 모델을 구성하였다. 대상하도는 금강유역의 용담댐과 대청댐 사이의 본류이며, 상류단 입력자료로써 본류에 있는 수통, 호탄 관측소 관측수위와 지류인 송천 관측소 관측수위를 고려하였다. 출력 값으로는 하류단의 옥천 관측소 수위를 3시간 및 6시간의 선행시간으로 예측하도록 인공신경망 모형을 구성하였다. 인공신경망의 학습(testing), 시험(testing), 검증(validation)을 위해 2000년부터 2012년까지 13년간의 시수위자료를 이용하여 학습을 진행하였으며, 2013년부터 2014년의 2년간의 수위자료를 이용한 시험을 통해 최적의 모형을 선정하였다. 또한 선정된 최적의 모형을 이용하여 2015년부터 2016년까지의 수위예측을 수행하였다.

• Journal of Korean Society of Steel Construction
• /
• v.19 no.6
• /
• pp.695-702
• /
• 2007

The Thermal Prestressing Method for continuous composite girder bridges is a new design and construction method developed to induce initial composite stresses in the concrete slab at negative bending regions. Due to the induced initial stresses, prevention of tensile cracks at the concrete slab, reduction of steel girder section, and reduction of reinforcing bars are possible. Thus, the construction efficiency can be improved and the construction can be made more economical. The method for determining the optimum heating region of the thermal prestressing method has not been established although such method is essential for improving the efficiency of the design process. The trial-and-error method used in previous studies is far from efficient, and a more rational method for computing optimal heating region is required. In this study, an efficient method for determining the optimum heating region in using the thermal prestressing method was developed based on the neural network algorithm, which is widely adopted to pattern recognition, optimization, diagnosis, and estimation problems in various fields. Back-propagation algorithm, commonly used as a learning algorithm in neural network problems, was used for the training of the neural network. Through case studies of two-span and three-span continuous composite girder bridges using the developed procedure, the optimal heating regions were obtained.

• Journal of Intelligence and Information Systems
• /
• v.22 no.2
• /
• pp.127-142
• /
• 2016

Deep learning model is a kind of neural networks that allows multiple hidden layers. There are various deep learning architectures such as convolutional neural networks, deep belief networks and recurrent neural networks. Those have been applied to fields like computer vision, automatic speech recognition, natural language processing, audio recognition and bioinformatics where they have been shown to produce state-of-the-art results on various tasks. Among those architectures, convolutional neural networks and recurrent neural networks are classified as the supervised learning model. And in recent years, those supervised learning models have gained more popularity than unsupervised learning models such as deep belief networks, because supervised learning models have shown fashionable applications in such fields mentioned above. Deep learning models can be trained with backpropagation algorithm. Backpropagation is an abbreviation for "backward propagation of errors" and a common method of training artificial neural networks used in conjunction with an optimization method such as gradient descent. The method calculates the gradient of an error function with respect to all the weights in the network. The gradient is fed to the optimization method which in turn uses it to update the weights, in an attempt to minimize the error function. Convolutional neural networks use a special architecture which is particularly well-adapted to classify images. Using this architecture makes convolutional networks fast to train. This, in turn, helps us train deep, muti-layer networks, which are very good at classifying images. These days, deep convolutional networks are used in most neural networks for image recognition. Convolutional neural networks use three basic ideas: local receptive fields, shared weights, and pooling. By local receptive fields, we mean that each neuron in the first(or any) hidden layer will be connected to a small region of the input(or previous layer's) neurons. Shared weights mean that we're going to use the same weights and bias for each of the local receptive field. This means that all the neurons in the hidden layer detect exactly the same feature, just at different locations in the input image. In addition to the convolutional layers just described, convolutional neural networks also contain pooling layers. Pooling layers are usually used immediately after convolutional layers. What the pooling layers do is to simplify the information in the output from the convolutional layer. Recent convolutional network architectures have 10 to 20 hidden layers and billions of connections between units. Training deep learning networks has taken weeks several years ago, but thanks to progress in GPU and algorithm enhancement, training time has reduced to several hours. Neural networks with time-varying behavior are known as recurrent neural networks or RNNs. A recurrent neural network is a class of artificial neural network where connections between units form a directed cycle. This creates an internal state of the network which allows it to exhibit dynamic temporal behavior. Unlike feedforward neural networks, RNNs can use their internal memory to process arbitrary sequences of inputs. Early RNN models turned out to be very difficult to train, harder even than deep feedforward networks. The reason is the unstable gradient problem such as vanishing gradient and exploding gradient. The gradient can get smaller and smaller as it is propagated back through layers. This makes learning in early layers extremely slow. The problem actually gets worse in RNNs, since gradients aren't just propagated backward through layers, they're propagated backward through time. If the network runs for a long time, that can make the gradient extremely unstable and hard to learn from. It has been possible to incorporate an idea known as long short-term memory units (LSTMs) into RNNs. LSTMs make it much easier to get good results when training RNNs, and many recent papers make use of LSTMs or related ideas.

• The Transactions of the Korea Information Processing Society
• /
• v.1 no.2
• /
• pp.225-236
• /
• 1994

Many research result conducted by neural network researchers claimed that the degree of generalization of the neural network system is higher or at least equal to that of statistical methods. However, those successful results could be brought only if the neural network was trained by appropriately sound data, having a little of noisy data and being large enough to control noisy data. Real data used in a lot of fields, especially business fields, were not so sound that the network have frequently failed to obtain satisfactory prediction accuracy, the degree of generalization. Enhancing the degree of generalization with noisy data is discussed in this study. The suggestion, which was obtained through a series of experiments, to enhance the degree of generalization is to remove inconsistent data by checking overlapping and inconsistencies. Furthermore, the previous conclusion by other reports is also confirmed that the learning mechanism of neural network takes average value of two inconsistent data included in training set［2］. The interim results of on-going research project are reported in this paper These are ann architecture of the neural network adopted in this project and the whole idea of developing on-line credit evaluation system,being intergration of the expert(resoning)system and the neural network(learning system.Another definite result is corroborated through this study that quickprop,being agopted as a learing algorithm, also has more speedy learning process than does back propagation even in very noisy environment.

• Computers and Concrete
• /
• v.19 no.5
• /
• pp.457-465
• /
• 2017

The Taiwanese liquid crystal display (LCD) industry has traditionally produced a huge amount of waste glass that is placed in landfills. Waste glass recycling can reduce the material costs of concrete and promote sustainable environmental protection activities. Concrete is always utilized as structural material; thus, the concrete compressive strength with a variety of mixtures must be studied using predictive models to achieve more precise results. To create an efficient waste LCD glass concrete (WLGC) design proportion, the related studies utilized a multivariable regression analysis to develop a compressive strength waste LCD glass concrete equation. The mix design proportion for waste LCD glass and the compressive strength relationship is complex and nonlinear. This results in a prediction weakness for the multivariable regression model during the initial growing phase of the compressive strength of waste LCD glass concrete. Thus, the R ratio for the predictive multivariable regression model is 0.96. Neural networks (NN) have a superior ability to handle nonlinear relationships between multiple variables by incorporating supervised learning. This study developed a multivariable prediction model for the determination of waste LCD glass concrete compressive strength by analyzing a series of laboratory test results and utilizing a neural network algorithm that was obtained in a related prior study. The current study also trained the prediction model for the compressive strength of waste LCD glass by calculating the effects of several types of factor combinations, such as the different number of input variables and the relevant filter for input variables. These types of factor combinations have been adjusted to enhance the predictive ability based on the training mechanism of the NN and the characteristics of waste LCD glass concrete. The selection priority of the input variable strategy is that evaluating relevance is better than adding dimensions for the NN prediction of the compressive strength of WLGC. The prediction ability of the model is examined using test results from the same data pool. The R ratio was determined to be approximately 0.996. Using the appropriate input variables from neural networks, the model validation results indicated that the model prediction attains greater accuracy than the multivariable regression model during the initial growing phase of compressive strength. Therefore, the neural-based predictive model for compressive strength promotes the application of waste LCD glass concrete.