• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.64-64
• /
• 2022

이상기후로 인해 돌발적이고 국지적인 호우 발생의 빈도가 증가하게 되면서 짧은 선행시간(~3 시간) 범위에서 수치예보보다 높은 정확도를 갖는 초단시간 강우예측자료가 돌발홍수 및 도시홍수의 조기경보를 위해 유용하게 사용되고 있다. 일반적으로 초단시간 강우예측 정보는 레이더를 활용하여 외삽 및 이동벡터 기반의 예측기법으로 산정한다. 최근에는 장기간 레이더 관측자료의 확보와 충분한 컴퓨터 연산자원으로 인해 레이더 자료를 활용한 인공지능 심층학습 기반(RNN(Recurrent Neural Network), CNN(Convolutional Neural Network), Conv-LSTM 등)의 강우예측이 국외에서 확대되고 있고, 국내에서도 ConvLSTM 등을 활용한 연구들이 진행되었다. CNN 심층신경망 기반의 초단기 예측 모델의 경우 대체적으로 외삽기반의 예측성능보다 우수한 경향이 있었으나, 예측시간이 길어질수록 공간 평활화되는 경향이 크게 나타나므로 고강도의 뚜렷한 강수 특징을 예측하기 힘들어 예측정확도를 향상시키는데 중요한 소규모 기상현상을 왜곡하게 된다. 본 연구에서는 이러한 한계를 보완하기 위해 적대적 생성 신경망(Generative Adversarial Network, GAN)을 적용한 초단시간 예측기법을 활용하고자 한다. GAN은 생성모형과 판별모형이라는 두 신경망이 서로간의 적대적인 경쟁을 통해 학습하는 신경망으로, 데이터의 확률분포를 학습하고 학습된 분포에서 샘플을 쉽게 생성할 수 있는 기법이다. 본 연구에서는 2017년부터 2021년까지의 환경부 대형 강우레이더 합성장을 수집하고, 강우발생 사례를 대상으로 학습을 수행하여 신경망을 최적화하고자 한다. 학습된 신경망으로 강우예측을 수행하여, 국내 기상청과 환경부에서 생산한 레이더 초단시간 예측강우와 정량적인 정확도를 비교평가 하고자 한다.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.20 no.3
• /
• pp.171-179
• /
• 2020

Recently, studies have shown that convolution neural networks are achieving the best performance in image classification, object detection, and image generation. Vision based defect inspection which is more economical than other defect inspection, is a very important for a factory automation. Although supervised anomaly detection algorithm has far exceeded the performance of traditional machine learning based method, it is inefficient for real industrial field due to its tedious annotation work, In this paper, we propose ADGAN, a unsupervised anomaly detection architecture using the variational autoencoder and the generative adversarial network which give great results in image generation task, and demonstrate whether the proposed network architecture identifies anomalous images well on MNIST benchmark dataset as well as our own welding defect dataset.

• The Journal of the Acoustical Society of Korea
• /
• v.38 no.6
• /
• pp.670-677
• /
• 2019

As the presence of background noise in acoustic signal degrades the performance of speech or acoustic event recognition, it is still challenging to extract noise-robust acoustic features from noisy signal. In this paper, we propose a combined structure of Wasserstein Generative Adversarial Network (WGAN) and MultiTask AutoEncoder (MTAE) as deep learning architecture that integrates the strength of MTAE and WGAN respectively such that it estimates not only noise but also speech features from noisy acoustic source. The proposed MTAE-WGAN structure is used to estimate speech signal and the residual noise by employing a gradient penalty and a weight initialization method for Leaky Rectified Linear Unit (LReLU) and Parametric ReLU (PReLU). The proposed MTAE-WGAN structure with the adopted gradient penalty loss function enhances the speech features and subsequently achieve substantial Phoneme Error Rate (PER) improvements over the stand-alone Deep Denoising Autoencoder (DDAE), MTAE, Redundant Convolutional Encoder-Decoder (R-CED) and Recurrent MTAE (RMTAE) models for robust speech recognition.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.4
• /
• pp.591-597
• /
• 2022

Recently, many studies have been done on the data augmentation technique as a way to efficiently build datasets. Among them, a representative data augmentation technique is a method of utilizing Generative Adversarial Network (GAN), which generates data similar to real data by competitively learning generators and discriminators. However, when learning GAN, there are cases where a broken pixel image occurs among similar data generated according to the environment and progress, which cannot be used as a dataset and causes an increase in learning time. In this paper, an algorithm was developed to select these damaged images by analyzing the histogram of image data generated during the GAN learning process, and as a result of comparing them with the images generated in the existing GAN, the ratio of the damaged images was reduced by 33.3 times(3,330%).

• Journal of IKEEE
• /
• v.25 no.4
• /
• pp.766-769
• /
• 2021

In order to perform tasks such as design, control, optimization, and prediction of flight vehicle trajectories based on machine learning techniques including deep learning, a certain amount of flight vehicle trajectory data is required. However, there are cases in which it is difficult to secure more than a certain amount of flight vehicle trajectory data for various reasons. In such cases, synthetic data generation could be one way to make machine learning possible. In this paper, to explore this possibility, we generated and evaluated synthetic flight vehicle trajectory data using time-series generative adversarial neural network. In addition, various ablation studies (comparative experiments) were performed to explore the possibility of using synthetic data in the aircraft trajectory prediction task. The experimental results presented in this paper are expected to be of practical help to researchers who want to conduct research on the possibility of using synthetic data in the generation of synthetic flight vehicle trajectory data and the work related to flight vehicle trajectories.

• Journal of Technologic Dentistry
• /
• v.46 no.2
• /
• pp.36-41
• /
• 2024

Purpose: This study aimed to generate virtual mandibular left first molar teeth using deep convolutional generative adversarial networks (DCGANs) and analyze their matching accuracy with actual tooth morphology to propose a new paradigm for using medical data. Methods: Occlusal surface images of the mandibular left first molar scanned using a dental model scanner were analyzed using DCGANs. Overall, 100 training sets comprising 50 original and 50 background-removed images were created, thus generating 1,000 virtual teeth. These virtual teeth were classified based on the number of cusps and occlusal surface ratio, and subsequently, were analyzed for consistency by expert dental technicians over three rounds of examination. Statistical analysis was conducted using IBM SPSS Statistics ver. 23.0 (IBM), including intraclass correlation coefficient for intrarater reliability, one-way ANOVA, and Tukey's post-hoc analysis. Results: Virtual mandibular left first molars exhibited high consistency in the occlusal surface ratio but varied in other criteria. Moreover, consistency was the highest in the occlusal buccal lingual criteria at 91.9%, whereas discrepancies were observed most in the occusal buccal cusp criteria at 85.5%. Significant differences were observed among all groups (p<0.05). Conclusion: Based on the classification of the virtually generated left mandibular first molar according to several criteria, DCGANs can generate virtual data highly similar to real data. Thus, subsequent research in the dental field, including the development of improved neural network structures, is necessary.

• Korean Journal of Radiology
• /
• v.23 no.1
• /
• pp.139-149
• /
• 2022

Objective: To compare the effects of bone suppression imaging using deep learning (BSp-DL) based on a generative adversarial network (GAN) and bone subtraction imaging using a dual energy technique (BSt-DE) on radiologists' performance for pulmonary nodule detection on chest radiographs (CXRs). Materials and Methods: A total of 111 adults, including 49 patients with 83 pulmonary nodules, who underwent both CXR using the dual energy technique and chest CT, were enrolled. Using CT as a reference, two independent radiologists evaluated CXR images for the presence or absence of pulmonary nodules in three reading sessions (standard CXR, BSt-DE CXR, and BSp-DL CXR). Person-wise and nodule-wise performances were assessed using receiver-operating characteristic (ROC) and alternative free-response ROC (AFROC) curve analyses, respectively. Subgroup analyses based on nodule size, location, and the presence of overlapping bones were performed. Results: BSt-DE with an area under the AFROC curve (AUAFROC) of 0.996 and 0.976 for readers 1 and 2, respectively, and BSp-DL with AUAFROC of 0.981 and 0.958, respectively, showed better nodule-wise performance than standard CXR (AUAFROC of 0.907 and 0.808, respectively; p ≤ 0.005). In the person-wise analysis, BSp-DL with an area under the ROC curve (AUROC) of 0.984 and 0.931 for readers 1 and 2, respectively, showed better performance than standard CXR (AUROC of 0.915 and 0.798, respectively; p ≤ 0.011) and comparable performance to BSt-DE (AUROC of 0.988 and 0.974; p ≥ 0.064). BSt-DE and BSp-DL were superior to standard CXR for detecting nodules overlapping with bones (p < 0.017) or in the upper/middle lung zone (p < 0.017). BSt-DE was superior (p < 0.017) to BSp-DL in detecting peripheral and sub-centimeter nodules. Conclusion: BSp-DL (GAN-based bone suppression) showed comparable performance to BSt-DE and can improve radiologists' performance in detecting pulmonary nodules on CXRs. Nevertheless, for better delineation of small and peripheral nodules, further technical improvements are required.

• Journal of Information Processing Systems
• /
• v.20 no.3
• /
• pp.375-390
• /
• 2024

Edge computing architecture has effectively alleviated the computing pressure on cloud platforms, reduced network bandwidth consumption, and improved the quality of service for user experience; however, it has also introduced new security issues. Existing anomaly detection methods in big data scenarios with cloud-edge computing collaboration face several challenges, such as sample imbalance, difficulty in dealing with complex network traffic attacks, and difficulty in effectively training large-scale data or overly complex deep-learning network models. A lightweight deep-learning model was proposed to address these challenges. First, normalization on the user side was used to preprocess the traffic data. On the edge side, a trained Wasserstein generative adversarial network (WGAN) was used to supplement the data samples, which effectively alleviates the imbalance issue of a few types of samples while occupying a small amount of edge-computing resources. Finally, a trained lightweight deep learning network model is deployed on the edge side, and the preprocessed and expanded local data are used to fine-tune the trained model. This ensures that the data of each edge node are more consistent with the local characteristics, effectively improving the system's detection ability. In the designed lightweight deep learning network model, two sets of convolutional pooling layers of convolutional neural networks (CNN) were used to extract spatial features. The bidirectional long short-term memory network (BiLSTM) was used to collect time sequence features, and the weight of traffic features was adjusted through the attention mechanism, improving the model's ability to identify abnormal traffic features. The proposed model was experimentally demonstrated using the NSL-KDD, UNSW-NB15, and CIC-ISD2018 datasets. The accuracies of the proposed model on the three datasets were as high as 0.974, 0.925, and 0.953, respectively, showing superior accuracy to other comparative models. The proposed lightweight deep learning network model has good application prospects for anomaly traffic detection in cloud-edge collaborative computing architectures.

• International Journal of Contents
• /
• v.16 no.2
• /
• pp.8-18
• /
• 2020

Data is critical in deep learning but the scarcity of data often occurs in research, especially in the preparation of the paired training data. In this paper, document image binarization with unpaired data is studied by introducing adversarial learning, excluding the need for supervised or labeled datasets. However, the simple extension of the previous unpaired training to binarization inevitably leads to poor performance compared to paired data training. Thus, a new deep learning approach is proposed by introducing a multi-diversity of higher quality generated images. In this paper, a two-stage model is proposed that comprises the generative adversarial network (GAN) followed by the U-net network. In the first stage, the GAN uses the unpaired image data to create paired image data. With the second stage, the generated paired image data are passed through the U-net network for binarization. Thus, the trained U-net becomes the binarization model during the testing. The proposed model has been evaluated over the publicly available DIBCO dataset and it outperforms other techniques on unpaired training data. The paper shows the potential of using unpaired data for binarization, for the first time in the literature, which can be further improved to replace paired data training for binarization in the future.

• Journal of Korea Multimedia Society
• /
• v.20 no.8
• /
• pp.1208-1215
• /
• 2017

Recently, people's attention and worries about fine particulate matter have been increasing. Due to the construction and maintenance costs, there are insufficient air quality monitoring stations. As a result, people have limited information about the concentration of fine particulate matter, depending on the location. Studies have been undertaken to estimate the fine particle concentrations in areas without a measurement station. Yet there are limitations in that the estimate cannot take account of other factors that affect the concentration of fine particle. In order to solve these problems, we propose a framework for estimating the concentration of fine particulate matter of a specific area using meteorological data and traffic data. Since there are more grids without a monitor station than grids with a monitor station, we used a domain adversarial neural network based on the domain adaptation method. The features extracted from meteorological data and traffic data are learned in the network, and the air quality index of the corresponding area is then predicted by the generated model. Experimental results demonstrate that the proposed method performs better as the number of source data increases than the method using conditional random fields.