• The Journal of Society for e-Business Studies
• /
• 제22권2호
• /
• pp.169-185
• /
• 2017

Juliano Rizzo and Thai Duong, the authors of the BEAST attack [11, 12] on SSL, have proposed a new attack named CRIME [13] which is Compression Ratio Info-leak Made Easy. The CRIME exploits how data compression and encryption interact to discover secret information about the underlying encrypted data. Repeating this method allows an attacker to eventually decrypt the data and recover HTTP session cookies. This security weakness targets in SPDY and SSL/TLS compression. The attack becomes effective because the attacker is enable to choose different input data and observe the length of the encrypted data that comes out. Since Transport Layer Security (TLS) ensures integrity of data transmitted between two parties (server and client) and provides strong authentication for both parties, in the last few years, it has a wide range of attacks on SSL/TLS which have exploited various features in the TLS mechanism. In this paper, we will discuss about the CRIME and other versions of SSL/TLS attacks along with countermeasures, implementations. We also present direction for SSL/TLS attacks resistance in practice.

• Journal of the Korean Data and Information Science Society
• /
• 제27권1호
• /
• pp.67-74
• /
• 2016

Due to the character of the climate of Korea, the pavement of a road is Influenced by freezing in winter season and thawing in thawing season. In the last few years, several articles have been devoted to the study to minimize the damage of freezing and thawing action. The purpose of this paper is to identify appropriacy of factors that influence road pavement thickness. We conduct the decision tree analysis on the field data of road pavement. The target variable is 'Frost penetration'. This value was calculated from the temperature data. The input variables are 'Region', 'Type of road pavement', 'Anti-frost layer', 'Month' and 'Air temperature'. The region was divided into 9 regions by freezing index $350{\sim}450^{\circ}C{\cdot}day$, $450{\sim}550^{\circ}C{\cdot}day$, $550{\sim}650^{\circ}C{\cdot}day$. The type of road pavement has three-section such as area of cutting, boundary area of cutting and bankin, lower area of banking. As the result, the variables that influence 'Frost penetration' are Month, followed by anti-frost layer, air temperature and region.

• Geophysics and Geophysical Exploration
• /
• 제19권2호
• /
• pp.76-83
• /
• 2016

Migration image can be distorted due to reflected waves in the source and receiver wavefields when discontinuities of input velocity model exist in seismic imaging. To remove reflected waves coming from layer interfaces, it is a common practice to smooth the velocity model for migration. If the velocity model is smoothed, however, the subsurface image can be distorted because the velocity changes around interfaces. In this paper, we attempt to minimize the distortion by reducing reflection energy in the source and receiver wavefields through acoustic impedance homogenization. To make acoustic impedance constant, we define fake density model and use it for migration. When the acoustic impedance is constant over all layers, the reflection coefficient at normal incidence becomes zero and the minimized reflection energy results in the improvement of migration result. To verify our algorithm, we implement the reverse-time migration using cell-based finite-difference method. Through numerical examples, we can note that the migration image is improved at the layer interfaces with high velocity contrast, and it shows the marked improvement particularly in the shallow part.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• 제11권2호
• /
• pp.204-208
• /
• 2018

In this paper, we proposed and evaluated the time series deep learning prediction model for learning fluctuation pattern of stock price. Recurrent neural networks, which can store previous information in the hidden layer, are suitable for the stock price prediction model, which is time series data. In order to maintain the long - term dependency by solving the gradient vanish problem in the recurrent neural network, we use LSTM with small memory inside the recurrent neural network. Furthermore, we proposed the stock price prediction model using bidirectional LSTM recurrent neural network in which the hidden layer is added in the reverse direction of the data flow for solving the limitation of the tendency of learning only based on the immediately preceding pattern of the recurrent neural network. In this experiment, we used the Tensorflow to learn the proposed stock price prediction model with stock price and trading volume input. In order to evaluate the performance of the stock price prediction, the mean square root error between the real stock price and the predicted stock price was obtained. As a result, the stock price prediction model using bidirectional LSTM recurrent neural network has improved prediction accuracy compared with unidirectional LSTM recurrent neural network.

• Journal of the Korean Society of Radiology
• /
• 제15권2호
• /
• pp.153-158
• /
• 2021

X-ray image analysis is a very important field to improve the early diagnosis rate and prediction accuracy of periodontal disease. Research on the development and application of artificial intelligence-based algorithms to improve the quality of such dental X-ray images is being widely conducted worldwide. Thus, the aim of this study was to design a super-resolution algorithm for predicting periodontal disease and to evaluate its applicability in dental X-ray images. The super-resolution algorithm was constructed based on the convolution layer and ReLU, and an image obtained by up-sampling a low-resolution image by 2 times was used as an input data. Also, 1,500 dental X-ray data used for deep learning training were used. Quantitative evaluation of images used root mean square error and structural similarity, which are factors that can measure similarity through comparison of two images. In addition, the recently developed no-reference based natural image quality evaluator and blind/referenceless image spatial quality evaluator were additionally analyzed. According to the results, we confirmed that the average similarity and no-reference-based evaluation values were improved by 1.86 and 2.14 times, respectively, compared to the existing bicubic-based upsampling method when the proposed method was used. In conclusion, the super-resolution algorithm for predicting periodontal disease proved useful in dental X-ray images, and it is expected to be highly applicable in various fields in the future.

• Journal of Korea Water Resources Association
• /
• 제51권12호
• /
• pp.1217-1227
• /
• 2018

This paper aims to evaluate the applicability of dam inflow prediction model using recurrent neural network theory. To achieve this goal, the Artificial Neural Network (ANN) model and the Elman Recurrent Neural Network(RNN) model were applied to hydro-meteorological data sets for the Soyanggang dam and the Chungju dam basin during dam operation period. For the model training, inflow, rainfall, temperature, sunshine duration, wind speed were used as input data and daily inflow of dam for 10 days were used for output data. The verification was carried out through dam inflow prediction between July, 2016 and June, 2018. The results showed that there was no significant difference in prediction performance between ANN model and the Elman RNN model in the Soyanggang dam basin but the prediction results of the Elman RNN model are comparatively superior to those of the ANN model in the Chungju dam basin. Consequently, the Elman RNN prediction performance is expected to be similar to or better than the ANN model. The prediction performance of Elman RNN was notable during the low dam inflow period. The performance of the multiple hidden layer structure of Elman RNN looks more effective in prediction than that of a single hidden layer structure.

• Journal of Climate Change Research
• /
• 제1권2호
• /
• pp.147-161
• /
• 2010

The temporal and spatial distributions of surface solar radiation were calculated by the one layer solar radiative transfer model(GWNU) which was corrected by multi layer Line-by-Line(LBL) model during 2009 in South Korea. The aerosol optical thickness, ozone amount, cloud fraction and total precipitable water were used as the input data for GWNU model run and they were retrieved from Moderate Resolution Imaging Spectrometer(MODIS), Ozone Monitoring Instrument(OMI), MTSAT-1R satellite data and the Regional Data Assimilation Prediction System(RDAPS) model result, respectively. The surface solar radiation was calculated with 4 km spatial resolution in South Korea region using the GWNU model and the results were compared with surface measurement(by pyranometer) data of 22 KMA solar sites. The maximum values(more than $5,400MJ/m^2$) of model calculated annual solar radiation were found in Andong, Daegu and Jinju regions and these results were corresponded with the MTSAT-1R cloud amount data. However, the spatial distribution of surface measurement data was comparatively different from the model calculation because of the insufficient correction and management problems for the sites instruments(pyranometer).

• Journal of IKEEE
• /
• 제25권2호
• /
• pp.273-279
• /
• 2021

In this paper, we propose an algorithm that shortens the learning time by performing individual learning using partitioning the deep learning structure. The proposed algorithm consists of four processes: network classification origin setting process, feature vector extraction process, feature noise removal process, and class classification process. First, in the process of setting the network classification starting point, the division starting point of the network structure for effective feature vector extraction is set. Second, in the feature vector extraction process, feature vectors are extracted without additional learning using the weights previously learned. Third, in the feature noise removal process, the extracted feature vector is received and the output value of each class is learned to remove noise from the data. Fourth, in the class classification process, the noise-removed feature vector is input to the multi-layer perceptron structure, and the result is output and learned. To evaluate the performance of the proposed algorithm, we experimented with the Extended Yale B face database. As a result of the experiment, in the case of the time required for one-time learning, the proposed algorithm reduced 40.7% based on the existing algorithm. In addition, the number of learning up to the target recognition rate was shortened compared with the existing algorithm. Through the experimental results, it was confirmed that the one-time learning time and the total learning time were reduced and improved over the existing algorithm.

• Journal of the Korean Geotechnical Society
• /
• 제38권11호
• /
• pp.149-161
• /
• 2022

This study selected the indexes for evaluating the damage of the railway embankments due to liquefaction from the earthquake damage cases of railway embankments. The study correlated the selected indexes and the settlement of the embankment crest from the dynamic numerical analysis. Further, the correlation was used to develop a method for evaluating the liquefaction damage to the railway embankment. The damage cases and damage types were analyzed, and referring to the liquefaction damage assessment method for other structures, the embankment height (H), the non-liquefiable layer thickness (H₁), and the liquefaction potential index were selected as indexes for evaluating the damage. The study performed dynamic effective stress analyses on the railway embankment, and the PM4-Sand model was applied as the constitutive liquefaction model for the embankment foundation ground. The model's validity was first verified by comparing it with the existing dynamic centrifugal model test results performed on the railway embankment. Nine sites where the foundation ground can be liquefied were selected from the data of 549 embankments of the Honam High-speed Railway in Korea. Further, dynamic numerical analyses using four seismic waves as input earthquake load were performed for the selected site sections. The numerical analysis results confirmed the correlation between the evaluation indexes and the embankment crest settlement. A method for efficiently evaluating the damage to the embankment due to liquefaction was proposed using the chart obtained from this correlation.

• Korean Journal of Optics and Photonics
• /
• 제34권4호
• /
• pp.157-169
• /
• 2023

Augmented reality (AR) using a head mounted display (HMD) is used in various fields such as military, medicine, manufacturing, gaming, and education. In this paper, we discuss the design and fabrication of the AR optical system, which is most essential for HMD. The AR optical system for HMD requires a wide transparent area in which the augmented image of the display and the real world can be viewed at the same time. To this end, an AR optical system was designed and manufactured by dividing it into three parts according to each characteristic. Also, the refractive index of the ultra-violet (UV) adhesive layer required to make the three optical systems into one complete AR optical system was considered from the design stage to minimize the optical path shift phenomenon when the input light source passes through the UV adhesive layer. In addition, when designing the AR optical system, two aspheric surfaces were used to compensate for off-axis aberration and to be suitable for mass production. Finally, for HMD mass production, an aspheric AR optical system with a thickness of 11 mm, a diagonal field of view of 40°, and a weight of 11.3 g was designed and manufactured.