• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.465-469
• /
• 2019

자연어 처리를 위해서 두 문장의 의미 유사성을 분석하는 것은 아주 중요하다. 이 논문은 패러프레이즈 검출 태스크를 수행하기 위한 Paraphrase-BERT를 제안한다. 우선 구글이 제안한 사전 학습된 BERT를 그대로 이용해서 패러프레이즈 데이터 (MRPC)를 가지고 파인 튜닝하였고 추가적으로 최근에 구글에서 새로 발표한 Whole Word Masking 기술을 사용하여 사전 학습된 BERT 모델을 새롭게 파인 튜닝하였다. 그리고 마지막으로 다중 작업 학습을 수행하여 성능을 향상시켰다. 구체적으로 질의 응답 태스크와 패러프레이즈 검출 태스크를 동시에 학습하여 후자가 더 잘 수행될 수 있도록 하였다. 결과적으로 점점 더 성능이 개선되었고 (11.11%의 정확도 향상, 7.88%의 F1 점수 향상), 향후 작업으로 파인 튜닝하는 방법에 대해서 추가적으로 연구할 계획이다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.574-576
• /
• 2021

본 논문에서는 딥러닝을 이용한 구글 ML Kit 를 이용하여 직접적이고 효과적인 졸음운전 예방기술을 구현하였다. 본 연구에서는 눈 상태를 인식하여 졸음을 감지하고 경보음을 발생시켜 교통사고 안전성 향상을 위한 방안을 제안하고 구현하였다. 또한, 정부 공공데이터 활용을 통해 성능테스트를 진행하여 시스템의 성능을 검증하였다.

• Proceedings of the Korean Society of Marine Engineers Conference
• /
• 2011.06a
• /
• pp.294-294
• /
• 2011

• Journal of the Korean association of regional geographers
• /
• v.17 no.1
• /
• pp.122-139
• /
• 2011

Yanbian area is a region with great interests to Cultural-Historically Korea as well as China. Cultural-historic study on Yanbian are lots of researches but can find few mapping for sites. This study aimed to make a map and analyze characteristics of distributions in the Cultural-Historic sites of Yanbian using Google Earth. We made a distribution map from stone age to Qing Dynasty. Symbology for mapping made color symbols by time series and categorical symbols. As a research finding, Sites of Balhae and Yuo-Geum age account for large parts in comparison with other ages in Yanbian. Especially, sites of Goguryeo, Balhae and Yuo-Geum age showed spatio-temporal structure of accumulative layers Characteristics of distributions is located in basin and stream area in the early age, and after then historical period moved to hilly sides and mountainous areas. The result of this research is expected to offer information for relevant follow-up studies of Cultural-Historic sites.

• Journal of Navigation and Port Research
• /
• v.35 no.1
• /
• pp.9-15
• /
• 2011

IMO adopt the revision of SOLAS convention on requirement systems for ECDIS and considered an ECDIS as the major system for E-Navigation strategy on marine transportation safety and environment protection. ENC(Electronic Navigational Chart) as the base map of ECDIS is considered as a principal information infrastructure that is essential for navigation tasks. Because ENCs are not easy to utilize for being encoded according to ISO/IEC 8211 file format, used for navigational purpose mainly, it's required to utilize in parts of Marine GIS and various marine application. In this paper, we developed KML conversion technology for ENC application. In details, we analyzed data model and encoding method of ENC, schema structure and data file of KML. Based on this results, we developed the program of KML conversion of ENCs.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.10
• /
• pp.13-20
• /
• 2013

Currently, SEAMCAT has been widely used as a tool to evaluate the effects of interference among wireless communication systems. In the previous work, we have incorporated the ITU-R P.526 pathloss model to the existing SEAMCAT in order to support the capability of interference evaluation taking into account any specific terrain characteristics. Along with this, we have implemented a terrain display function based on the Google map. However, the two-dimensional Google map based display is not effective in helping users to figure out some terrain features including the elevation variation in a given region. In order to alleviate this difficulty, we have incorporated the three-dimensional terrain display using the API of the Google earth to the existing SEAMCAT and provided the capability of viewing the positions of the associated communication systems, the variation of the carrier intensity and interference intensity in location, shadow region indication, and line-of-sight analysis and presented an example of interference evaluation.

• Journal of Digital Convergence
• /
• v.18 no.5
• /
• pp.417-423
• /
• 2020

The purpose of this study is to study the elements that make up the interface design of the applicatio, focusing on Android's Google Material Design Guidelines. and propose a way to produce the UI design guide by categorizing it according to form and combination methods. The components provided by the Materials Design Guidelines were disassembled form and attributes, types, and the characteristics. The components of UI design include surfaces, letters, icons, and media, which enable the creation of UI style guides, and the components for the purpose of exploring and communicating information, with different components and rules of use used for the purpose. Based on these results, I would like to propose design criteria for the interface of mobile applications and how to utilize it effectively.

• Science of Emotion and Sensibility
• /
• v.23 no.1
• /
• pp.69-78
• /
• 2020

This study compares and analyzes user experiences of intelligent personal assistant services based on the evaluation criteria of human-computer interaction to explore positive elements of user experiences and factors that could be improved. The research was conducted on Apple's Siri, Google's Google Associate, and Samsung's Bixby, which is presently the smartest personal assistant service on the market. The research method was to compare and analyze the concepts and characteristics of the current services through a literature review and by interviewing seven UI/UX design experts for the second 2 weeks using contextual inquiry. The interview reorganized Peter Morville's user experience honeycomb, reducing his seven usability principles down to five, asking questions about usability, convenience, visual attractiveness, reliability, and satisfaction. On the basis of the reconfigured usability principle, the assessment was conducted on the basis of the assessment timing and the system usability scale. This study is meaningful in that it analyzed the user experience of artificial intelligence personal assistant services both quantitatively and qualitatively.

• The Journal of Korean Association of Computer Education
• /
• v.14 no.4
• /
• pp.11-20
• /
• 2011

Recently, the interests of mobile GIS(Geographic Information System) technology is increasing with the spread of wireless network, the improvement of mobile device's performances and the growth of demands about mobile services. If we refer mobile GIS system classified as 3rd generation with the success of Web 2.0, it is suggested that supplies of information from the server side couldn't enhance users' participations and the usage of service. In this study, we suggested the model of GIS 2.0 application based on Google map for Android smart-phones, which users could make, save and share information using photos and text by themselves. We were able to grasp the expectation of GIS 2.0 by expert assessment of this application and got to the core of conditions and implications for the success of GIS 2.0.

• Smart Media Journal
• /
• v.9 no.3
• /
• pp.41-45
• /
• 2020

Recently, artificial intelligence (AI) and the Internet of things (IoT), which are represented by machine learning and deep learning among IT technologies related to the Fourth Industrial Revolution, are applied to our real life in various fields through various researches. In this paper, IoT and AI using object recognition technology are applied to classify clothing. For this purpose, the image dataset was taken using webcam and raspberry pi, and GoogLeNet, a convolutional neural network artificial intelligence network, was applied to transfer the photographed image data. The clothing image dataset was classified into two categories (shirtwaist, trousers): 900 clean images, 900 loss images, and total 1800 images. The classification measurement results showed that the accuracy of the clean clothing image was about 97.78%. In conclusion, the study confirmed the applicability of other objects using artificial intelligence networks on the Internet of Things based platform through the measurement results and the supplementation of more image data in the future.