• Journal of Biomedical Engineering Research
• /
• v.42 no.4
• /
• pp.150-158
• /
• 2021

Intracranial hemorrhage (ICH) refers to acute bleeding inside the intracranial vault. Not only does this devastating disease record a very high mortality rate, but it can also cause serious chronic impairment of sensory, motor, and cognitive functions. Therefore, a prompt and professional diagnosis of the disease is highly critical. Noninvasive brain imaging data are essential for clinicians to efficiently diagnose the locus of brain lesion, volume of bleeding, and subsequent cortical damage, and to take clinical interventions. In particular, computed tomography (CT) images are used most often for the diagnosis of ICH. In order to diagnose ICH through CT images, not only medical specialists with a sufficient number of diagnosis experiences are required, but even when this condition is met, there are many cases where bleeding cannot be successfully detected due to factors such as low signal ratio and artifacts of the image itself. In addition, discrepancies between interpretations or even misinterpretations might exist causing critical clinical consequences. To resolve these clinical problems, we developed a diagnostic model predicting intracranial bleeding and its subtypes (intraparenchymal, intraventricular, subarachnoid, subdural, and epidural) by applying deep learning algorithms to CT images. We also constructed a visualization tool highlighting important regions in a CT image for predicting ICH. Specifically, 1) 27,758 CT brain images from RSNA were pre-processed to minimize the computational load. 2) Three different CNN-based models (ResNet, EfficientNet-B2, and EfficientNet-B7) were trained based on a training image data set. 3) Diagnosis performance of each of the three models was evaluated based on an independent test image data set: As a result of the model comparison, EfficientNet-B7's performance (classification accuracy = 91%) was a way greater than the other models. 4) Finally, based on the result of EfficientNet-B7, we visualized the lesions of internal bleeding using the Grad-CAM. Our research suggests that artificial intelligence-based diagnostic systems can help diagnose and treat brain diseases resolving various problems in clinical situations.

• Journal of Internet Computing and Services
• /
• v.23 no.1
• /
• pp.77-86
• /
• 2022

The Republic of Korea is a country in which coastal surveillance is an imperative national task as it is surrounded by seas on three sides under the confrontation between South and North Korea. However, due to Defense Reform 2.0, the number of R/D (Radar) operating personnel has decreased, and the period of service has also been shortened. Moreover, there is always a possibility that a human error will occur. This paper presents specific guidelines for developing an AI learning model for the intelligent coastal surveillance system. We present a three-step strategy to realize the guidelines. The first stage is a typical stage of building an AI learning model, including data collection, storage, filtering, purification, and data transformation. In the second stage, R/D signal analysis is first performed. Subsequently, AI learning model development for classifying real and false images, coastal area analysis, and vulnerable area/time analysis are performed. In the final stage, validation, visualization, and demonstration of the AI learning model are performed. Through this research, the first achievement of making the existing weapon system intelligent by applying the application of AI technology was achieved.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.24 no.4
• /
• pp.341-353
• /
• 2022

Pipelines are buried in urban area, and the position (depth and orientation) of buried pipeline should be clearly identified before ground excavation. Although various geophysical methods can be used to detect the buried pipeline, it is not easy to identify the exact information of pipeline due to heterogeneous ground condition. Among various non-destructive geo-exploration methods, ground penetration radar (GPR) can explore the ground subsurface rapidly with relatively low cost compared to other exploration methods. However, the exploration data obtained from GPR requires considerable experiences because interpretation is not intuitive. Recently, researches on automated detection technology for GPR data using deep learning have been conducted. However, the lack of GPR data which is essential for training makes it difficult to build up the reliable detection model. To overcome this problem, we conducted a preliminary study to improve the performance of the detection model using finite difference time domain (FDTD)-based numerical analysis. Firstly, numerical analysis was performed with homogeneous soil media having single permittivity. In case of heterogeneous ground, numerical analysis was performed considering the ground heterogeneity using fractal technique. Secondly, deep learning was carried out using convolutional neural network. Detection Model-A is trained with data set obtained from homogeneous ground. And, detection Model-B is trained with data set obtained from homogeneous ground and heterogeneous ground. As a result, it is found that the detection Model-B which is trained including heterogeneous ground shows better performance than detection Model-A. It indicates the ground heterogeneity should be considered to increase the performance of automated detection model for GPR exploration.

• The Korean Journal of Applied Statistics
• /
• v.36 no.2
• /
• pp.101-114
• /
• 2023

The data augmentation technique is effectively used to solve the problem of overfitting the model by allowing the training dataset to be viewed from various perspectives. In addition to image augmentation techniques such as rotation, cropping, horizontal flip, and vertical flip, occlusion-based data augmentation methods such as Cutmix and Cutout have been proposed. For models based on speech data, it is possible to use an occlusion-based data-based augmentation technique after converting a 1D speech signal into a 2D spectrogram. In particular, SpecAugment is an occlusion-based augmentation technique for speech spectrograms. In this study, we intend to compare and study data augmentation techniques that can be used in the problem of false-voice detection. Using data from the ASVspoof2017 and ASVspoof2019 competitions held to detect fake audio, a dataset applied with Cutout, Cutmix, and SpecAugment, an occlusion-based data augmentation method, was trained through an LCNN model. All three augmentation techniques, Cutout, Cutmix, and SpecAugment, generally improved the performance of the model. In ASVspoof2017, Cutmix, in ASVspoof2019 LA, Mixup, and in ASVspoof2019 PA, SpecAugment showed the best performance. In addition, increasing the number of masks for SpecAugment helps to improve performance. In conclusion, it is understood that the appropriate augmentation technique differs depending on the situation and data.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.4
• /
• pp.329-344
• /
• 2023

In this study, we comprehensively analyze the generalization performance of various deep learning-based active sonar target classifiers when applied to small and imbalanced active sonar datasets. To generate the active sonar datasets, we use data from two different oceanic experiments conducted at different times and ocean. Each sample in the active sonar datasets is a time-frequency domain image, which is extracted from audio signal of contact after the detection process. For the comprehensive analysis, we utilize 22 Convolutional Neural Networks (CNN) models. Two datasets are used as train/validation datasets and test datasets, alternatively. To calculate the variance in the output of the target classifiers, the train/validation/test datasets are repeated 10 times. Hyperparameters for training are optimized using Bayesian optimization. The results demonstrate that shallow CNN models show superior robustness and generalization performance compared to most of deep CNN models. The results from this paper can serve as a valuable reference for future research directions in deep learning-based active sonar target classification.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.29 no.7
• /
• pp.836-842
• /
• 2023

Under the Welfare of Disabled Persons Act, the national qualification test allows disabled and non-disabled people to appear for the same test, and the deaf do not fall under the reason for disqualification from obtaining a license for power-driven water leisure crafts; therefore, even the deaf can obtain a license. During a risk of collision at sea, ships and power-driven water leisure crafts notify the other party of dangerous signals such as maneuvering and warning signals through sound signals to prevent accidents that may occur because surrounding ships are not visible. However, a method is required to prevent marine accidents that may occur when the deaf cannot hear danger signals through sounds from nearby ships or power-driven water leisure crafts during leisure activities owing to hearing impairment. A sound reception system is a device installed on a ship where missionary work is completely deposed to display the amplification of external sounds and the direction of sound reception on the screen. Through visual display of sound signals such as dangerous signals that cannot be heard owing to hearing impairment, improvement measures were proposed to prevent marine accidents that can occur owing to the inability of the deaf to check sound signals during leisure activities.

• The Journal of the Acoustical Society of Korea
• /
• v.29 no.4
• /
• pp.278-285
• /
• 2010

The difference between training and test environments is one of the major performance degradation factors in noisy speech recognition and many silence feature normalization methods were proposed to solve this inconsistency. Conventional silence feature normalization method represents higher classification performance in higher SNR, but it has a problem of performance degradation in low SNR due to the low accuracy of speech/silence classification. On the other hand, cepstral distance represents well the characteristic distribution of speech/silence (or noise) in low SNR. In this paper, we propose a Cepstral distance and Log-energy based Silence Feature Normalization (CLSFN) method which uses both log-energy and cepstral euclidean distance to classify speech/silence for better performance. Because the proposed method reflects both the merit of log energy being less affected with noise in high SNR and the merit of cepstral distance having high discrimination accuracy for speech/silence classification in low SNR, the classification accuracy will be considered to be improved. The experimental results showed that our proposed CLSFN presented the improved recognition performances comparing with the conventional SFN-I/II and CSFN methods in all kinds of noisy environments.

• Journal of the Korea Institute of Construction Safety
• /
• v.6 no.1
• /
• pp.27-32
• /
• 2024

In this study, based on accident cases of manned and unmanned tower cranes, the types and causes of accidents were analyzed and tower crane accident prevention measures were proposed accordingly. As a result, 'lack of communication' was commonly analyzed as the biggest cause of accidents. Therefore, in this study, we investigated domestic and international systems and laws regarding signalmen responsible for communication between tower crane operators and workers and proposed the following three improvement methods. First, the signalman system must be legally and institutionally reestablished. Second, national agencies and private organizations with expertise should create standardized training manuals and distribute them to signalman educators, and create checklists to manage educators. Third, in the case of a site using a tower crane, the employer must mandatorily deploy a signalman with professional qualifications. Lastly, the results of this study are expected to highlight the differences in accident types and causes of manned and unmanned tower cranes and the need for legal and institutional improvement in the signal system.

• Journal of Korea Water Resources Association
• /
• v.49 no.7
• /
• pp.589-597
• /
• 2016

In this study, we investigated the use of satellite-derived flow (SDF) signals and a data-based model for the estimation of outflow for the river reach where in situ measurements are either completely unavailable or are difficult to access for hydraulic and hydrology analysis such as the upper basin of Imjin River. It has been demonstrated by many studies that the SDF signals can be used as the river width estimates and the correlation between SDF signals and river width is related to the shape of cross sections. To extract the nonlinear relationship between SDF signals and river outflow, Artificial Neural Network (ANN) model with SDF signals as its inputs were applied for the computation of flow discharge at Imjin Bridge located in Imjin River. 15 pixels were considered to extract SDF signals and Partial Mutual Information (PMI) algorithm was applied to identify the most relevant input variables among 150 candidate SDF signals (including 0~10 day lagged observations). The estimated discharges by ANN model were compared with the measured ones at Imjin Bridge gauging station and correlation coefficients of the training and validation were 0.86 and 0.72, respectively. It was found that if the 1 day previous discharge at Imjin bridge is considered as an input variable for ANN model, the correlation coefficients were improved to 0.90 and 0.83, respectively. Based on the results in this study, SDF signals along with some local measured data can play an useful role in river flow estimation and especially in flood forecasting for data-scarce regions as it can simulate the peak discharge and peak time of flood events with satisfactory accuracy.

• Journal of Intelligence and Information Systems
• /
• v.16 no.3
• /
• pp.77-97
• /
• 2010

Market timing is an investment strategy which is used for obtaining excessive return from financial market. In general, detection of market timing means determining when to buy and sell to get excess return from trading. In many market timing systems, trading rules have been used as an engine to generate signals for trade. On the other hand, some researchers proposed the rough set analysis as a proper tool for market timing because it does not generate a signal for trade when the pattern of the market is uncertain by using the control function. The data for the rough set analysis should be discretized of numeric value because the rough set only accepts categorical data for analysis. Discretization searches for proper "cuts" for numeric data that determine intervals. All values that lie within each interval are transformed into same value. In general, there are four methods for data discretization in rough set analysis including equal frequency scaling, expert's knowledge-based discretization, minimum entropy scaling, and na$\ddot{i}$ve and Boolean reasoning-based discretization. Equal frequency scaling fixes a number of intervals and examines the histogram of each variable, then determines cuts so that approximately the same number of samples fall into each of the intervals. Expert's knowledge-based discretization determines cuts according to knowledge of domain experts through literature review or interview with experts. Minimum entropy scaling implements the algorithm based on recursively partitioning the value set of each variable so that a local measure of entropy is optimized. Na$\ddot{i}$ve and Booleanreasoning-based discretization searches categorical values by using Na$\ddot{i}$ve scaling the data, then finds the optimized dicretization thresholds through Boolean reasoning. Although the rough set analysis is promising for market timing, there is little research on the impact of the various data discretization methods on performance from trading using the rough set analysis. In this study, we compare stock market timing models using rough set analysis with various data discretization methods. The research data used in this study are the KOSPI 200 from May 1996 to October 1998. KOSPI 200 is the underlying index of the KOSPI 200 futures which is the first derivative instrument in the Korean stock market. The KOSPI 200 is a market value weighted index which consists of 200 stocks selected by criteria on liquidity and their status in corresponding industry including manufacturing, construction, communication, electricity and gas, distribution and services, and financing. The total number of samples is 660 trading days. In addition, this study uses popular technical indicators as independent variables. The experimental results show that the most profitable method for the training sample is the na$\ddot{i}$ve and Boolean reasoning but the expert's knowledge-based discretization is the most profitable method for the validation sample. In addition, the expert's knowledge-based discretization produced robust performance for both of training and validation sample. We also compared rough set analysis and decision tree. This study experimented C4.5 for the comparison purpose. The results show that rough set analysis with expert's knowledge-based discretization produced more profitable rules than C4.5.