• Journal of Positioning, Navigation, and Timing
• /
• v.9 no.3
• /
• pp.151-156
• /
• 2020

A technology for calculating the position of a device is very important for users who receive positioning services, regardless of various indoor/outdoor or with/without any positioning infrastructure existence environments. One of the positioning resources widely used at present, LTE, is a typical infrastructure that can overcome the space limitation, however its positioning method based on the position of the LTE base station has low accuracy. A method of constructing a radio wave map of an LTE signal has been proposed as a method for overcoming the accuracy, but it takes a lot of time and cost to perform high-density collection in a wide area. In this paper, we describe a method of creating a high-density DB for the entire region by using vehicle-based partial collection data. To create a positioning database, we applied the idea of Generative Adversarial Network (GAN), which has recently been in the spotlight in the field of deep learning, and learned the collected data. Then, a virtually generated map which having the smallest error from the actual data is selected as the optimum DB. We verified the effectiveness of the positioning DB generation algorithm using the positioning data obtained from un-collected area.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.12
• /
• pp.4326-4344
• /
• 2021

The SinGAN is one of generative adversarial networks that can be trained on a single nature image. It has poor ability to learn more global features from nature image, and losses much local detail information when it generates arbitrary size image sample. To solve the problem, a non-linear function is firstly proposed to control downsampling ratio that is ratio between the size of current image and the size of next downsampled image, to increase the ratio with increase of the number of downsampling. This makes the low-resolution images obtained by downsampling have higher proportion in all downsampled images. The low-resolution images usually contain much global information. Therefore, it can help the model to learn more global feature information from downsampled images. Secondly, the attention mechanism is introduced to the generative network to increase the weight of effective image information. This can make the network learn more local details. Besides, in order to make the output image more natural, the TVLoss function is introduced to the loss function of SinGAN, to reduce the difference between adjacent pixels and smear phenomenon for the output image. A large number of experimental results show that our proposed model has better performance than other methods in generating random samples with fixed size and arbitrary size, image harmonization and editing.

• Journal of Internet Computing and Services
• /
• v.21 no.2
• /
• pp.121-130
• /
• 2020

This paper suggests evaluating the military closed network data as an image which is generated by Generative Adversarial Network (GAN), applying an image evaluation method such as the InceptionV3 model-based Inception Score (IS) and Frechet Inception Distance (FID). We employed the famous image classification models instead of the InceptionV3, added layers to those models, and converted the network data to an image in diverse ways. Experimental results show that the Densenet121 model with one added Dense Layer achieves the best performance in data converted using the arctangent algorithm and 8 * 8 size of the image.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.4
• /
• pp.1795-1811
• /
• 2019

In order to achieve rapid and accurate detection of vehicle objects in complex traffic conditions, we propose a novel vehicle detection method. Firstly, more contextual and small-object vehicle information can be obtained by our Joint Feature Network (JFN). Secondly, our Evolved Region Proposal Network (EPRN) generates initial anchor boxes by adding an improved version of the region proposal network in this network, and at the same time filters out a large number of false vehicle boxes by soft-Non Maximum Suppression (NMS). Then, our Mask Network (MaskN) generates an example that includes the vehicle occlusion, the generator and discriminator can learn from each other in order to further improve the vehicle object detection capability. Finally, these candidate vehicle detection boxes are optimized to obtain the final vehicle detection boxes by the Fine-Tuning Network(FTN). Through the evaluation experiment on the DETRAC benchmark dataset, we find that in terms of mAP, our method exceeds Faster-RCNN by 11.15%, YOLO by 11.88%, and EB by 1.64%. Besides, our algorithm also has achieved top2 comaring with MS-CNN, YOLO-v3, RefineNet, RetinaNet, Faster-rcnn, DSSD and YOLO-v2 of vehicle category in KITTI dataset.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.10
• /
• pp.67-76
• /
• 2023

Although deep learning models are making innovative achievements in the field of computer vision, the problem of vulnerability to adversarial examples continues to be raised. Adversarial examples are attack methods that inject fine noise into images to induce misclassification, which can pose a serious threat to the application of deep learning models in the real world. In this paper, we propose a model that detects adversarial examples using differences in predictive values between edge-learned classification models and underlying classification models. The simple process of extracting the edges of the objects and reflecting them in learning can increase the robustness of the classification model, and economical and efficient detection is possible by detecting adversarial examples through differences in predictions between models. In our experiments, the general model showed accuracy of {49.9%, 29.84%, 18.46%, 4.95%, 3.36%} for adversarial examples (eps={0.02, 0.05, 0.1, 0.2, 0.3}), whereas the Canny edge model showed accuracy of {82.58%, 65.96%, 46.71%, 24.94%, 13.41%} and other edge models showed a similar level of accuracy also, indicating that the edge model was more robust against adversarial examples. In addition, adversarial example detection using differences in predictions between models revealed detection rates of {85.47%, 84.64%, 91.44%, 95.47%, and 87.61%} for each epsilon-specific adversarial example. It is expected that this study will contribute to improving the reliability of deep learning models in related research and application industries such as medical, autonomous driving, security, and national defense.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2019.10a
• /
• pp.879-881
• /
• 2019

전자의무기록 (EMR)과 같은 의료 현장에서 수집되는 대용량의 데이터는 임상 해석적으로 잠재가치가 크고 활용도가 다양하나 결측값이 많아 희소성이 크다는 한계점이 있어 분석이 어렵다. 특히 EMR의 정보수집과정에서 발생하는 결측값은 무작위적이고 임의적이어서 분석 정확도를 낮추고 예측 모델의 성능을 저하시키는 주된 요인으로 작용하기 때문에, 결측치 대체는 필수불가결하다. 최근 통상적으로 활용되어지던 통계기반 알고리즘기반의 결측치 대체 알고리즘보다는 딥러닝 기술을 활용한 알고리즘들이 새로이 등장하고 있다. 본 논문에서는 Generative Adversarial Network를 기반한 최신 결측값 대치 알고리즘인 Generative Adversarial Imputation Nets을 적용하여 EMR에서의 성능을 분석해보고자 하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.513-514
• /
• 2023

초해상은 저해상도의 영상을 고해상도 영상으로 합성하는 기술이다. 이 기술에 딥러닝이 적용되어, 2014년에는 SRCNN(Super Resolution Convolutional Neural Network) 모델이 발표됐다. 이후에는 SRCAE(Super Resolution Convolutional Autoencoders)와 GAN(Generative Adversarial Networks)을 기반으로 한 SRGAN(Super Resolution Generative Adversarial Networks) 등, SRCNN의 성능을 능가하는 모델들이 발표됐다. ESRGAN(Enhanced Super Resolution Generative Adversarial Networks)은 SRGAN 모델의 성능을 개선했지만, 완벽한 성능을 내지 못하는 문제점이 있다. 이에 본 논문에서는 판별자(Discriminator) 구조를 변경하여 ESRGAN의 성능을 개선한다. 실험 결과, 제안하는 모델이 ESRGAN보다 더 높은 성능을 보일 것으로 기대된다.

• The Journal of Korea Robotics Society
• /
• v.13 no.1
• /
• pp.26-30
• /
• 2018

Planetary global localization is necessary for long-range rover missions in which communication with command center operator is throttled due to the long distance. There has been number of researches that address this problem by exploiting and matching rover surroundings with global digital elevation maps (DEM). Using conventional methods for matching, however, is challenging due to artifacts in both DEM rendered images, and/or rover 2D images caused by DEM low resolution, rover image illumination variations and small terrain features. In this work, we use train CNN discriminator to match rover 2D image with DEM rendered images using conditional Generative Adversarial Network architecture (cGAN). We then use this discriminator to search an uncertainty bound given by visual odometry (VO) error bound to estimate rover optimal location and orientation. We demonstrate our network capability to learn to translate rover image into DEM simulated image and match them using Devon Island dataset. The experimental results show that our proposed approach achieves ~74% mean average precision.

• KIPS Transactions on Software and Data Engineering
• /
• v.8 no.6
• /
• pp.257-264
• /
• 2019

Recently, deep neural network based approaches have shown a good performance for various fields of natural language processing. A huge amount of training data is essential for building a deep neural network model. However, collecting a large size of training data is a costly and time-consuming job. A data augmentation is one of the solutions to this problem. The data augmentation of text data is more difficult than that of image data because texts consist of tokens with discrete values. Generative adversarial networks (GANs) are widely used for image generation. In this work, we generate sentimental texts by using one of the GANs, CS-GAN model that has a discriminator as well as a classifier. We evaluate the usefulness of generated sentimental texts according to various measurements. CS-GAN model not only can generate texts with more diversity but also can improve the performance of its classifier.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.38 no.6
• /
• pp.499-509
• /
• 2020

This study explores how to build object classification learning data based on artificial intelligence. The data has been investigated recently in image classification fields and, in turn, has a great potential to use. In order to recognize and extract relatively accurate objects using artificial intelligence, a large amount of learning data is required to be used in artificial intelligence algorithms. However, currently, there are not enough datasets for object recognition learning to share and utilize. In addition, generating data requires long hours of work, high expenses and labor. Therefore, in the present study, a small amount of initial aerial image learning data was used in the GAN (Generative Adversarial Network)-based generator network in order to establish image learning data. Moreover, the experiment also evaluated its quality in order to utilize additional learning datasets. The method of oversampling learning data using GAN can complement the amount of learning data, which have a crucial influence on deep learning data. As a result, this method is expected to be effective particularly with insufficient initial datasets.