• Journal of Digital Contents Society
• /
• v.18 no.5
• /
• pp.969-976
• /
• 2017

This paper proposes the learning method of an artificial neural network and a convolutional neural network using the WFSO algorithm developed as an optimization algorithm. Since the optimization algorithm searches based on a number of candidate solutions, it has a drawback in that it is generally slow, but it rarely falls into the local optimal solution and it is easy to parallelize. In addition, the artificial neural networks with non-differentiable activation functions can be trained and the structure and weights can be optimized at the same time. In this paper, we describe how to apply WFSO algorithm to artificial neural network learning and compare its performances with error back-propagation algorithm in multilayer artificial neural networks and convolutional neural networks.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2011.05a
• /
• pp.371-371
• /
• 2011

본 연구에서는 한강유역 내 관측기간이 충분한 기상청 지상관측소 10개소를 선정하고 CCCma(Canadian Century for Climate modeling and analysis)에서 제공하는 자료에 대한 인공신경망기법 상세화 적용을 실시하였다. 인공신경망의 학습을 위해 CGCM3.1/T63 20C3M시나리오(reference scenario)의 22개 2D변수 중 물리적으로 민감도가 높다고 판단되는 GCM_Prec, huss, ps를 입력변수로 선정하였으며 인공신경망 학습기간은 1991년~1995년, 검증기간은 1996년~2000년, 예측기간은 2011년~2100년으로 A1B, A2 B1 시나리오 등 다양한 기후변화 시나리오를 통해 예측band를 제시하고자 하였다. 하지만 공간상관을 고려하기 위하여 각 관측소에 대하여 인공신경망 학습을 하는 경우 관측소간 spatial correlation 및 spatial cluster구현이 어렵기 때문에 Spatial Rectangular Pulse모형을 이용하고자 하였으나, 강수면적에 대한 scale의 결정이 어렵다는 단점을 확인 하고 본 연구에서는 Random Cascade 모형을 이용하여 ${\beta}$를 통한 강수면적 scale(rainy area fraction)을 결정하고자 하였다. Random Cascade모형의 기법은 격자단위의 downscaling기법으로 강수대의 공간적 형상을 재현하며 스케일에 비종속적인(scale-invariant)프랙탈 특성을 이용하여 매개변수를 최소화 할 수 있는 장점을 가진 기법으로 한강유역 1Km내외 강우장을 만들어 topographic effect를 첨가하고자 한다.

• Journal of Digital Contents Society
• /
• v.17 no.6
• /
• pp.449-459
• /
• 2016

This paper proposes a new method to improve the three problems of existing automatic composition method using artificial neural networks. The first problem is that the existing beat post-processing to fit into music theories could not handle all the cases of occurring. The second one is that the pitch space generated by artificial neural networks is distorted because the rest is trained with the pitch on the same neural network with large values. The last problem is caused by the difference between the initial melody and beats given by user and those generated by an artificial neural network in the process of new composition. In order to treat these problems, we propose an enhanced post-processing of beats, initial melody processing, and adoption of artificial neural network for rest. It was found from experiments that the proposed methods totally resolved the three problems.

• Journal of Digital Contents Society
• /
• v.18 no.4
• /
• pp.641-647
• /
• 2017

In this paper, we propose a method to automatically generate a configured song with melodies composed of front/middle/last parts by using hierarchical artificial neural networks in automatic composition. In the first layer, an artificial neural network is used to learn an existing song or a random melody and outputs a song after performing rhythm post-processing. In the second layer, the melody created by the artificial neural network in the first layer is learned by three artificial neural networks of front/middle/last parts in the second layer in order to make a configured song. In the artificial neural network of the second layer, we applied a method to generate repeatability using measure identity in order to make song with repeatability and after that the song is completed after rhythm, chord, tonality post-processing. It was confirmed from experiments that our proposed method produced configured songs well.

• Journal of the Korean GEO-environmental Society
• /
• v.13 no.8
• /
• pp.25-33
• /
• 2012

Although it is important to reflect the accurate information of the ground condition in the tunnel design, the analysis and design are conducted by limited information because it is very difficult to consider various geographies and geotechnical conditions. When the tunnel is under construction, examination of accurate safety and prediction of behavior are overcome the limits of predicting behavior by Artificial Neural Network in this study. First, construct the suitable structure after the data of field was made sure by the multi-layer back propagation, then apply with algorithm. Employ the result of measured data from database, and consider the influence factor of tunnel, like supporting pattern, RMR, Q, the types of rock, excavation length, excavation shape, excavation over, to carry out the reliable analysis through field applicability of Artificial Neural Network. After studying, using the ANN model to predict the shearing displacement, convergence displacement, underground displacement, Rock bolt output follow the excavation over of tunnel construction field, then determine the field applicability with ANN through field measured value and comparison analysis when tunnel is being constructed.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.21 no.1
• /
• pp.54-61
• /
• 2001

A software package to classify acoustic emission (AE) signals using the wavelet transform and the neural network was developed Both of the continuous and the discrete wavelet transforms are considered, and the error back-propagation neural network is adopted as m artificial neural network algorithm. The signals acquired during the 3-point bending test of specimens which have artificial defects on weld zone are used for the classification of the defects. Features are extracted from the time-frequency plane which is the result of the wavelet transform of signals, and the neural network classifier is tamed using the extracted features to classify the signals. It has been shown that the developed software package is useful to classify AE signals. The difference between the classification results by the continuous and the discrete wavelet transforms is also discussed.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.26 no.6
• /
• pp.445-451
• /
• 2016

This paper introduces a postprocessing method, an iteration method for melody, and an average neural network method for learning a large number of songs in order to improve musically insufficient parts in automatic composition using existing artificial neural network. The melody of songs composed by artificial neural networks is produced according to the melodies of trained songs, so it can not be a specific tonality and it is difficult to have a repetitive composition. In order to solve these problems, we propose a postprocessing method that converts the melody composed by artificial neural networks into a melody having a specific tonality according to music theory and an iteration method for melody by iteratively composing measure divisions of artificial neural networks. In addition, the existing training method of many songs has some disadvantages. To solve this problem, we adopt an average neural network that is made by averaging the weights of artificial neural networks trained each song. From some experiments, it was confirmed that the proposed method solves the existing problems.

• Journal of the Korean Geotechnical Society
• /
• v.19 no.4
• /
• pp.287-298
• /
• 2003

The feasibility of using neural networks to model the complex relationship between piezocone measurements and the undrained shear strength of clays has been investigated. A three layered back propagation neural network model was developed based on actual undrained shear strengths, which were obtained from the isotrpoically and anisotrpoically consolidated triaxial compression test(CIUC and CAUC), and piezocone measurements compiled from various locations around the world. It was validated by comparing model predictions with measured values about new piezocone data, which were not previously employed during development of model. Performance of the neural network model was compared with conventional empirical method, direct correlation method, and theoretical method. It was found that the neural network model is not only capable of inferring a complex relationship between piezocone measurements and the undrained shear strength of clays but also gives a more precise and reliable undrained shear strength than theoretical and empirical approaches. Furthermore, neural network model has a possibility to be a generalized relationship between piezocone measurements and undrained shear strength over the various places and countries, while the present empirical correlations present the site specific relationship.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2017.05a
• /
• pp.120-120
• /
• 2017

인천지역의 상수도공급은 팔당댐을 취수원으로 하여 도수, 송수관을 거쳐 인천지역 내 정수장을 통하여 각 급수지역까지 일원화된 관로시스템으로 공급되고 있다. 관망에서의 적절한 수압관리, 노후관로 교체사업 등은 급수관망 내 관로 사고위험을 줄일 수 있고, 누수량을 저감하여 무수율의 감소로 이어질 수 있다. 상수관망 내 누수에 영향을 주는 물리적, 운영적 요소를 파악하고, 이를 이용하여 누수해결을 위한 방법론을 제시하는 것은 매우 중요하다. 본 연구에서는 인천시 배수관망 데이터를 활용하여 통계분석 및 인공신경망을 통하여 무수율에 영향을 미치는 인자를 선별하고, 무수율과의 연관성을 분석하고자 하였다. 이를 위해 대상지역에 대한 시설현황 및 운영자료를 취득하고, 무수율 분석에 활용하였다. 인천시의 소블럭을 대상으로 관로노후도, 배수관연장, 평균관경, 급수전당 공급량, 누수발생 횟수, 용도지역, 관망구성 형태 등을 고려하여 무수율과의 관계분석을 위한 통계분석을 수행하였다. 특히 급수에 필요한 최소에너지와 관망에서 공급되는 에너지를 비교하기 위하여 관망해석 프로그램인 EPANET을 이용하여 관망내 절점에서의 수압과 수요량이 적용된 최소공급에너지를 활용하였고, 이를 통하여 블록 내 과잉공급에너지와 무수율의 영향성을 비교하였다. 최종적으로 산출된 주요인자에 대한 주성분분석, 분산분석, 다중회귀분석 등의 통계분석과 인공신경망에 의해 학습된 알고리즘을 통하여 산정된 무수율을 실측 무수율과 비교, 분석하였다. 인공신경망에 의해 산정된 무수율과 실측 무수율의 정확도를 평가하기 위하여 MAE, MSE, PBIAS 등의 정확도 평가와 산점도 분석을 수행하고, 상관계수를 도출하여 가장 정확한 방법을 결정하였다. 분석 결과 통계분석에 의한 다중회귀식으로 산출된 무수율 보다 인공신경망에 의한 무수율이 실측값에 더욱 근접한 것으로 나타났으며 이용된 뉴런의 수의 따라 산출결과가 상이하기 때문에 최적 뉴런의 수를 산정해야 할 필요가 있음을 확인하였다. 특히 사용된 상수관망 주요인자 중 주성분분석을 통하여 선정된 각 성분을 인공신경망에 적용시 더욱 정확한 무수율 예측이 가능한 것으로 나타났다.

• Journal of the Korean Institute of Electrical and Electronic Material Engineers
• /
• v.35 no.2
• /
• pp.134-142
• /
• 2022

As industry and technology go through advancement, it is hard to search new materials which satisfy various standards through conventional trial-and-error based research methods. Crystal Graph Convolutional Neural Network(CGCNN) is a neural network which uses material's features as train data, and predicts the material properties(formation energy, bandgap, etc.) much faster than first-principles calculation. This report introduces how to train the CGCNN model which predicts the formation energy using open database. It is anticipated that with a simple programming skill, readers could construct a model using their data and purpose. Developing machine learning model for materials science is going to help researchers who should explore large chemical and structural space to discover materials efficiently.