• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 한국정보통신학회 2022년도 춘계학술대회
• /
• pp.377-380
• /
• 2022

An application called an app can be downloaded and used on mobile devices. Among them, Android-based apps have the disadvantage of being implemented on an open source basis and can be exploited by anyone, but unlike iOS, which discloses only a small part of the source code, Android is implemented as an open source, so it can analyze the code. However, since anyone can participate in changing the source code of open source-based Android apps, the number of malicious codes increases and types are bound to vary. Malicious codes that increase exponentially in a short period of time are difficult for humans to detect one by one, so it is efficient to use a technique to detect malicious codes using AI. Most of the existing malicious app detection methods are to extract Features and detect malicious apps. Therefore, three ways to select the optimal feature to be used for learning after feature extraction are proposed. Finally, in the step of modeling with optimal features, ensemble techniques are used in addition to a single model. Ensemble techniques have already shown results beyond the performance of a single model, as has been shown in several studies. Therefore, this paper presents a plan to select the optimal feature and implement a learning model for Android app-based malicious code detection.

• Journal of the Korean Society for information Management
• /
• 제40권2호
• /
• pp.115-135
• /
• 2023

The purpose of this study is to assess the effectiveness of using deep learning language models to extract references automatically and create a reference database for research reports in an efficient manner. Unlike academic journals, research reports present difficulties in automatically extracting references due to variations in formatting across institutions. In this study, we addressed this issue by introducing the task of separating references from non-reference phrases, in addition to the commonly used metadata extraction task for reference extraction. The study employed datasets that included various types of references, such as those from research reports of a particular institution, academic journals, and a combination of academic journal references and non-reference texts. Two deep learning language models, namely RoBERTa+CRF and ChatGPT, were compared to evaluate their performance in automatic extraction. They were used to extract metadata, categorize data types, and separate original text. The research findings showed that the deep learning language models were highly effective, achieving maximum F1-scores of 95.41% for metadata extraction and 98.91% for categorization of data types and separation of the original text. These results provide valuable insights into the use of deep learning language models and different types of datasets for constructing reference databases for research reports including both reference and non-reference texts.

• Korean Journal of Remote Sensing
• /
• 제39권5_2호
• /
• pp.849-858
• /
• 2023

Investigating disaster sites such as earthquakes and landslides involves significant risks due to potential secondary disasters like facility collapse. In situations where direct access is challenging, there is a need to develop methods for safely acquiring high-precision 3D disaster information using light detection and ranging (LiDAR) equipped drone survey systems. In this study, the feasibility of using drone LiDAR in disaster scenarios was examined, focusing on the collapse accident at Jeongja Bridge in Bundang-gu, Seongnam City, in April 2023. High-density point clouds for the accident bridge were collected, and the bridge's 3D terrain information was reconstructed and compared to the measurement performance of 10 ground control points. The results showed horizontal and vertical root mean square error values of 0.032 m and 0.055 m, respectively. Additionally, when compared to a point cloud generated using ground LiDAR for the same target area, a vertical difference of approximately 0.08 m was observed, but overall shapes showed minimal discrepancies. Moreover, in terms of overall data acquisition and processing time, drone LiDAR was found to be more efficient than ground LiDAR. Therefore, the use of drone LiDAR in disaster sites with significant risks allows for safe and rapid onsite investigations.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• 제33권6호
• /
• pp.1077-1086
• /
• 2023

The searchable encryption model allows to selectively search for encrypted data stored on a remote server. In a real-world scenarios, the model must be able to support multiple search keywords, multiple data owners/users. In this paper, these models are referred to as Multi Ranked Searchable Encryption model. However, at the time this paper was written, the proposed models use fully-trusted trapdoor centers, some of which assume that the connection between the user and the trapdoor center is secure, which is unlikely that such assumptions will be kept in real life. In order to improve the practicality and security of these searchable encryption models, this paper proposes a new Multi Ranked Searchable Encryption model which uses random keywords to protect search words requested by the data downloader from an honest-but-curious trapdoor center with an external attacker without the assumptions. The attacker cannot distinguish whether two different search requests contain the same search keywords. In addition, experiments demonstrate that the proposed model achieves reasonable performance, even considering the overhead caused by adding this protection process.

• Journal of the Society of Disaster Information
• /
• 제20권2호
• /
• pp.302-314
• /
• 2024

Purpose: This study proposes a fall detection model based on a top-down deep learning pose estimation model to automatically determine falls of multiple workers in an underground utility tunnel, and evaluates the performance of the proposed model. Method: A model is presented that combines fall discrimination rules with the results inferred from YOLOv8-pose, one of the top-down pose estimation models, and metrics of the model are evaluated for images of standing and falling two or fewer workers in the tunnel. The same process is also conducted for a bottom-up type of pose estimation model (OpenPose). In addition, due to dependency of the falling interference of the models on worker detection by YOLOv8-pose and OpenPose, metrics of the models for fall was not only investigated, but also for person. Result: For worker detection, both YOLOv8-pose and OpenPose models have F1-score of 0.88 and 0.71, respectively. However, for fall detection, the metrics were deteriorated to 0.71 and 0.23. The results of the OpenPose based model were due to partially detected worker body, and detected workers but fail to part them correctly. Conclusion: Use of top-down type of pose estimation models would be more effective way to detect fall of workers in the underground utility tunnel, with respect to joint recognition and partition between workers.

• Journal of Cadastre & Land InformatiX
• /
• 제54권1호
• /
• pp.19-32
• /
• 2024

This study examined the feasibility of image-based surveys by detecting objects in facilities and agricultural land using the YOLO algorithm based on drone images and comparing them with the land category by law. As a result of detecting objects through the YOLO algorithm, buildings showed a performance of detecting objects corresponding to 96.3% of the buildings provided in the existing digital map. In addition, the YOLO algorithm developed in this study detected 136 additional buildings that were not located in the digital map. Plastic greenhouses detected a total of 297 objects, but the detection rate was low for some plastic greenhouses for fruit trees. Also, agricultural land had the lowest detection rate. This result is because agricultural land has a larger area and irregular shape than buildings, so the accuracy is lower than buildings due to the inconsistency of training data. Therefore, segmentation detection, rather than box-shaped detection, is likely to be more effective for agricultural fields. Comparing the detected objects with the land category by law, it was analyzed that some buildings exist in agricultural and forest areas where it is difficult to locate buildings. It seems that it is necessary to link with administrative information to understand that these buildings are used illegally. Therefore, at the current level, it is possible to objectively determine the existence of buildings in fields where it is difficult to locate buildings.

• Journal of the Society of Disaster Information
• /
• 제20권1호
• /
• pp.69-81
• /
• 2024

Purpose: Recently, there has been a movement to guarantee the right to watch sports for the disabled. However, the sports stadium is designed without considering the wheelchair users, so the right to move in the stadium is not secured. Restrictions on the movement of the disabled make the evacuation vulnerable in an emergency. This study aims to develop a plan to ensure the safety of movement and evacuation of wheelchair users by conducting simulations targeting indoor sports stadiums. Method: The simulation was performed by constructing a scenario with the shape of the stands as a variable. The effect of the installation of wheelchair seats on evacuation was confirmed. Result: The results according to whether wheelchair seats are installed, the evacuation route of wheelchair movement, and whether wheelchair seats are separately arranged were compared. The impact of wheelchair seat installation on evacuation and its characteristics were derived. As a result, upward and separation seat was the most vulnerable to evacuation. Conclusion: A plan to secure evacuation performance was derived for the top floors of upward and separation seat. It is judged that the content can be use as a way to secure the safety of movement and evacuation of the disabled in sports stadiums.

• Therapeutic Science for Rehabilitation
• /
• 제13권2호
• /
• pp.69-84
• /
• 2024

Objective : This study aimed to categorize the lifestyles of older adults into two types - healthy and unhealthy, and use machine learning to identify the personal values that influence these lifestyles. Methods : This cross-sectional study targeting middle-aged and older adults (55 years and above) living in local communities in South Korea. Data were collected from 300 participants through online surveys. Lifestyle types were dichotomized by the Yonsei Lifestyle Profile (YLP)-Active, Balanced, Connected, and Diverse (ABCD) responses using latent profile analysis. Personal value information was collected using YLP-Values (YLP-V) and analyzed using machine learning to identify the relative importance of personal values on lifestyle types. Results : The lifestyle of older adults was categorized into healthy (48.87%) and unhealthy (51.13%). These two types showed the most significant difference in social relationship characteristics. Among the machine learning models used in this study, the support vector machine showed the highest classification performance, achieving 96% accuracy and 95% area under the receiver operating characteristic (ROC) curve. The model indicated that individuals who prioritized a healthy diet, sought health information, and engaged in hobbies or cultural activities were more likely to have a healthy lifestyle. Conclusion : This study suggests the need to encourage the expansion of social networks among older adults. Furthermore, it highlights the necessity to comprehensively intervene in individuals' perceptions and values that primarily influence lifestyle adherence.

• Journal of the Korea Society of Computer and Information
• /
• 제29권8호
• /
• pp.11-21
• /
• 2024

This study explores the use of virtual reality (VR) technology to identify and label key segments of the golf swing. To address the limitations of existing VR devices, we developed a platform to collect kinematic data from various VR devices using the OpenVR SDK (Software Development Kit) and SteamVR, and developed a semi-automated labeling technique to identify and label temporal changes in kinematic behavior through LSTM (Long Short-Term Memory)-based time series data analysis. The experiment consisted of 80 participants, 20 from each of the following age groups: teenage, young-adult, middle-aged, and elderly, collecting data from five swings each to build a total of 400 kinematic datasets. The proposed technique achieved consistently high accuracy (≥0.94) and F1 Score (≥0.95) across all age groups for the seven main phases of the golf swing. This work aims to lay the groundwork for segmenting exercise data and precisely assessing athletic performance on a segment-by-segment basis, thereby providing personalized feedback to individual users during future education and training.

• The Journal of the Acoustical Society of Korea
• /
• 제20권4호
• /
• pp.35-41
• /
• 2001

This paper describes Linear Discriminant Analysis and common vector extraction for speech recognition. Voice signal contains psychological and physiological properties of the speaker as well as dialect differences, acoustical environment effects, and phase differences. For these reasons, the same word spelled out by different speakers can be very different heard. This property of speech signal make it very difficult to extract common properties in the same speech class (word or phoneme). Linear algebra method like BT (Karhunen-Loeve Transformation) is generally used for common properties extraction In the speech signals, but common vector extraction which is suggested by M. Bilginer et at. is used in this paper. The method of M. Bilginer et al. extracts the optimized common vector from the speech signals used for training. And it has 100％ recognition accuracy in the trained data which is used for common vector extraction. In spite of these characteristics, the method has some drawback-we cannot use numbers of speech signal for training and the discriminant information among common vectors is not defined. This paper suggests advanced method which can reduce error rate by maximizing the discriminant information among common vectors. And novel method to normalize the size of common vector also added. The result shows improved performance of algorithm and better recognition accuracy of 2％ than conventional method.