필기 인식은 사람이 작성한 문서나 종이에 쓴 글자, 사진에 보이는 글자 등을 인식하는 기술이다. 대표적인 기술로는 OCR과 온라인 필기인식 기술이 있으며 OCR은 정자로 또박또박 쓴 글씨 인식률은 높지만 그렇지 않는 경우에는 인식률이 낮다. 온라인 필기인식 기술은 필기 입력순서와 사람의 필체의 차이에 따라 인식률이 확연하게 달랐다. 본 논문에서는 이러한 단점을 보완하고자 딥러닝을 이용하여 필기체 인식 시스템을 제안하고자 한다. 본 논문에서는 신경망 알고리즘 중 Convolutional Neural Network와 EMNIST 데이터 세트를 사용하여 학습 데이터를 설계하였고 Unity3D 게임엔진을 이용하여 전체적인 시스템을 구성하였다. 또한 본 논문에서는 CPU와 GPU 성능이 학습 결과에 영향을 미치는지 알아보기 위해 성능을 비교분석을 하였고, loss 값과 accuracy 결과에 큰 차이는 없었지만 학습 속도에는 최대 30배 정도 속도 차이가 났다. 마지막으로 실험을 통해 시스템 인식결과를 분석하였고, 문자와 숫자가 유사한 O, q, l과 같은 알파벳이나, 실험자가 글자를 다른 알파벳과 유사하게 보이게 필기하면 인식률이 낮았다. 본 논문에서 제안하는 시스템은 게임엔진을 사용하여 인공지능 시스템을 개발했기 때문에 프로세스 절차가 간략해졌고 호환성도 좋아졌다.

The incidence of stomach cancer has been found to be gradually decreasing; however, it remains one of the most frequently occurring malignant cancers in Korea. According to statistics of 2017, stomach cancer is the top cancer in men and the fourth most important cancer in women, necessitating methods for its early detection and treatment. Considerable research in the field of bioinformatics has been conducted in cancer studies, and bioinformatics approaches might help develop methods and models for its early prediction. We aimed to develop a classification method based on deep learning and demonstrate its application to gene expression data obtained from patients with stomach cancer. Data of 60,483 genes from 334 patients with stomach cancer in The Cancer Genome Atlas were evaluated by principal component analysis, heatmaps, and the convolutional neural network (CNN) algorithm. We combined the RNA-seq gene expression data with clinical data, searched candidate genes, and analyzed them using the CNN deep learning algorithm. We performed learning using the sample type and vital status of patients with stomach cancer and verified the results. We obtained an accuracy of 95.96% for sample type and 50.51% for vital status. Despite overfitting owing to the limited number of patients, relatively accurate results for sample type were obtained. This approach can be used to predict the prognosis of stomach cancer, which has many types and underlying causes.

There has been considerable recent interest in deep learning techniques for structural analysis and design. However, despite newer algorithms and more precise methods have been developed in the field of computer science, the recent effective deep learning techniques have not been applied to the damage detection topics. In this study, we have explored the structural damage detection method of truss structures using the state-of-the-art deep learning techniques. The deep neural networks are used to train knowledge of the patterns in the response of the undamaged and the damaged structures. A 31-bar planar truss are considered to show the capabilities of the deep learning techniques for identifying the single or multiple-structural damage. The frequency responses and the elasticity moduli of individual elements are used as input and output datasets, respectively. In all considered cases, the neural network can assess damage conditions with very good accuracy.

증발산은 순복사 에너지를 사용하여 잠열의 형태로 수증기를 대기 중으로 수송함으로써 지구에너지 순환에 있어 중요한 요소 중의 하나이며, 증발산량은 지표유출의 두 배 정도로서 지구 물 수지에서 차지하는 비중이 매우 크다. 증발산의 지상관측은 지점에 국한되기 때문에 공간연속면 상에서의 증발산량 산출을 위하여 격자형 기상자료와 위성자료를 이용한 모델링이 오랫동안 이루어져왔다. PM(Penman-Monteith) 방정식에 기초한 METRIC(Mapping Evapotranspiration with Internalized Calibration) 모델이나 PT(Priestley-Taylor) 방정식을 이용한 MS-PT(Modified Satellite-based Priestley-Taylor) 모델 등이 주로 사용되어 왔으나, 또 하나의 대안으로서 본 연구에서는 최근 부각되고 있는 딥러닝 기법인 DNN(deep neural network)을 이용한 증발산 모델링을 수행하였다. 은닉층 구조, 손실함수, 옵티마이저, 활성화함수, L1/L2 정규화, 드롭아웃 비율 등의 최적화 과정을 거쳐서 수립한 DNN 모델은 RMSE = 0.326mm/day, 상관계수 = 0.975의 매우 양호한 정확도를 나타내었다. 이는 DNN 최적화와 함께, 국지예보모델과 위성자료로부터 증발산 기작에 관여하는 인자들을 선택하여 입력자료로 적절히 사용하였기 때문이기도 하다. 향후과제로서 훈련자료의 종류와 양을 증가시켜서 DNN 모델을 보다 정교화하는 것은 반드시 필요하다고 사료된다.

In this study, an algorithm applying deep learning to the truss structures was proposed. Deep learning is a method of raising the accuracy of machine learning by creating a neural networks in a computer. Neural networks consist of input layers, hidden layers and output layers. Numerous studies have focused on the introduction of neural networks and performed under limited examples and conditions, but this study focused on two- and three-dimensional truss structures to prove the effectiveness of algorithms. and the training phase was divided into training model based on the dataset size and epochs. At these case, a specific data value was selected and the error rate was shown by comparing the actual data value with the predicted value, and the error rate decreases as the data set and the number of hidden layers increases. In consequence, it showed that it is possible to predict the result quickly and accurately without using a numerical analysis program when applying the deep learning technique to the field of structural analysis.

PURPOSES: In this study, algorithms were proposed for determining the crack condition of an asphalt pavement image using deep learning methods. METHODS: For the configuration of a deep learning network, the study used a Convolution Neural Network and You Only Look Once algorithms. To obtain input data for analysis, a camera was mounted on the bonnet of the vehicle to obtain images of asphalt pavement and to mark the ground-truth cracks in the asphalt pavement image. In addition, an algorithm suitable for the automatic determination function of Deep Learning was proposed in order to calculate the crack ratio and crack rating. RESULTS: The result of analysis showed that the recall rate of cracks in this system was higher from FPPW 5.0E-06 to 96.03%. Furthermore, the accuracy of the grading system was found to be 100%, enabling the determination of very accurate ratings. The rate of processing per image was 0.4448 seconds on average, and the real-time analysis of pavement images presented no problem because the assessment took place within a short time. CONCLUSIONS : Applying this system to the pavement management system is expected to reduce the time required in finishing work and to determine a quantitative crack rating.

Global climate change and increased international travel have affected the transmission of mosquito-borne diseases. In South Korea, uncommon diseases such as Dengue, chikungunya and Zika virus could be transmitted by potent mediator like Aedes albopictus. In order to cope with the risk of mosquito-borne diseases, rapid mosquito monitoring system is needed. Current mosquito monitoring procedures include installation of outdoor traps-mosquito collection-species classification-analysis of disease detection – upload of information to government research institutes – disease alert. In this process, species classification takes a lot of time, and if we reduce the time, we can cope with the disease outbreak more quickly. In this study, we developed automate species classification system target for 5 mosquito species (Culex pipiens, Cx. tritaeniorhynchus, Ae. albpictus, Ae. vexans, Anopheles spp.) disease vector live in South Korea. After modeling the morphology of each mosquito species, machine learning was carried out using DenseNet (Densely Connected Networks), one of the models of Artificial Neural Network. Using the learned model, we tested the classification of 5 species of mosquitoes and showed the accuracy from 97.35% to 99.48% at the maximum. Future research will focus on increasing the number of identifiable mosquito species and reducing the time spent on species classification. The autonomous classification of mosquito species using Deep Learning technology will contribute to the development of mosquito monitoring system and public health.