• 한국전자통신학회논문지
• /
• 제15권6호
• /
• pp.1187-1194
• /
• 2020

본 논문은 온라인 퀴즈 시스템에서 핵심인 문제은행 구축 자동화를 위한 Deep Quiz Cropping 기법을 제시했다. 이것은 문제지를 스캔한 그림 파일에서 개별문제에 대한 질의영역과 선다영역을 딥러닝 기반 검출기를 통해 검출하는 것과, 문제생성을 위해 질의영역과 선다영역을 짝지우고 영역오류를 수정하는 Box Coupling으로 이루어졌다. 문제지 및 시험지를 스캔한 영상파일에 Deep Quiz Coupling 기법을 적용한 다수의 실험에서 질의영역과 선다영역을 검출하는데 있어서 성공적인 결과를 도출했다.

• 한국컴퓨터정보학회논문지
• /
• 제27권1호
• /
• pp.77-82
• /
• 2022

본 연구는 장소 이미지를 수집하고 학습하여 사용자가 관심이 있어 하는 이미지의 장소를 예측하여 알려주는 서비스 개발을 목적으로 한다. 이미지 학습을 위한 이미지 데이터들은 크롤링 부분을 통해 수집되도록 설계되었다. 이미지 수집 이후 수집된 이미지들은 장소별로 라벨링 되어 CNN의 다양한 층을 통하여 학습된다. 각 층을 거칠 때마다 입력받은 학습 데이터는 최적화하여 특징 맵과의 비교를 반복하며 특정 장소 이미지의 특징 정보를 뽑아낸다. 충분한 학습 데이터가 쌓이면 다양한 장소 이미지들에 대해 예측이 가능하다. 학습 결과 모델의 정확도는 79.2로 높은 학습 정확도를 보였다.

• 반도체디스플레이기술학회지
• /
• 제19권4호
• /
• pp.126-129
• /
• 2020

We introduce a lifetime assessment technique using deep learning algorithm with complex electrical parameters such as resistivity, permittivity, impedance parameters as integrated indicators for predicting the degradation of the organic molecules. The evaluation system consists of fully automated in-situ measurement system and multiple layer perceptron learning system with five hidden layers and 1011 perceptra in each layer. Prediction accuracies are calculated and compared depending on the physical feature, learning hyperparameters. 62.5% of full time-series data are used for training and its prediction accuracy is estimated as r-square value of 0.99. Remaining 37.5% of the data are used for testing with prediction accuracy of 0.95. With k-fold cross-validation, the stability to the instantaneous changes in the measured data is also improved.

• 한국가시화정보학회지
• /
• 제19권2호
• /
• pp.41-47
• /
• 2021

River and stream are the important water supply source in our lives. Eutrophication causes excessive green algae growth including microcystis, which makes harmful to ecosystem and human health. Therefore, the water purification process to remove green algae is essential. In Korea, green algae alarm system exists depending on the concentration of green algae cells in river or stream. To maintain the growth amount under control, green algae monitoring system is being used. However, the unmanned, small and automatic monitoring system would be preferable. In this study, we developed the 3D printed device to measure the concentration of green algae cell using microfluidic droplet generator and deep learning. Deep learning network was trained by using transfer learning through pre-trained deep learning network. This newly developed microfluidic cell counter has sufficient accuracy to be possibly applicable to green algae alarm system.

• 한국소프트웨어감정평가학회 논문지
• /
• 제15권1호
• /
• pp.63-70
• /
• 2019

최근 딥 러닝(Deep leaning)을 이용한 이미지 인식 분야의 연구가 활발히 진행되고 있다. 본 연구에서는 육안으로 관찰하여 분류하기 어려운 암석을 이미지만으로 분류하기 위해 딥 러닝 오픈 소스 프레임워크인 Tensorflow 기반의 CNN모델을 사용하여 고등학교 교육과정에서 다루는 암석 18종(화성암 6종, 변성암 6종, 퇴적암 6종)의 이미지를 통해 암석을 분류하는 시스템을 제안한다. 암석의 이미지를 학습시켜 암석을 구별하는 분류기를 개발하여 분류 성능을 확인하였으며 최종적으로 구현한 모바일 어플리케이션을 통해 교실 내 학습 또는 현장체험학습 등에서 학생들의 학습 보조도구로서 사용할 수 있다.

• 반도체디스플레이기술학회지
• /
• 제23권1호
• /
• pp.110-116
• /
• 2024

Despite the continuous advancement of science and technology, fire accidents continue to occur without decreasing over time, so there is a constant need for a system that can accurately detect fires at an early stage. However, because most existing fire detection systems detect fire in the early stage of combustion when smoke is generated, rapid fire prevention actions may be delayed. Therefore we propose an early fire detection system that can perform early fire detection at a reasonable cost using LSTM, a deep learning model based on multi-gas sensors with high selectivity in the early stage of decomposition rather than the smoke generation stage. This system combines multiple gas sensors to achieve faster detection speeds than traditional sensors. In addition, through window sliding techniques and model light-weighting, the false alarm rate is low while maintaining the same high accuracy as existing deep learning. This shows that the proposed fire early detection system is a meaningful research in the disaster and engineering fields.

• 스마트미디어저널
• /
• 제12권11호
• /
• pp.36-47
• /
• 2023

주요 산업현장에서 설비를 제어하는 산업제어시스템(ICS, Industrial Control System)이 네트워크로 다른 시스템과 연결되는 사례가 증가하고 있다. 또한, 이러한 통합과 함께 한 번의 외부 침입이 전체 시스템 마비로 이루어질 수 있는 지능화된 공격의 발달로, 산업제어시스템에 대한 보안에 대한 위험성과 파급력이 증가하고 있어, 사이버 공격에 대한 보호 및 탐지 방안의 연구가 활발하게 진행되고 있으며, 비지도학습 형태의 딥러닝 모델이 많은 성과를 보여 딥러닝을 기반으로 한 이상(Anomaly) 탐지 기술이 많이 도입되고 있다. 어어, 본 연구에서는 딥러닝 모델에 전처리 방법론을 적용하여 시계열 데이터의 이상 탐지성능을 향상시키는 것에 중점을 두어, 그 결과 웨이블릿 변환(WT, Wavelet Transform) 기반 노이즈 제거 방법론이 딥러닝 기반 이상 탐지의 전처리 방법론으로 효과적임을 알 수 있었으며, 특히 센서에 대한 군집화(Clustering)를 통해 센서의 특성을 반영하여 Dual-Tree Complex 웨이블릿 변환을 차등적으로 적용하였을 때 사이버 공격의 탐지성능을 높이는 것에 가장 효과적임을 확인하였다.

• Journal of Information Processing Systems
• /
• 제15권5호
• /
• pp.1119-1130
• /
• 2019

With the wide spread of Social Network Services (SNS), fake news-which is a way of disguising false information as legitimate media-has become a big social issue. This paper proposes a deep learning architecture for detecting fake news that is written in Korean. Previous works proposed appropriate fake news detection models for English, but Korean has two issues that cannot apply existing models: Korean can be expressed in shorter sentences than English even with the same meaning; therefore, it is difficult to operate a deep neural network because of the feature scarcity for deep learning. Difficulty in semantic analysis due to morpheme ambiguity. We worked to resolve these issues by implementing a system using various convolutional neural network-based deep learning architectures and "Fasttext" which is a word-embedding model learned by syllable unit. After training and testing its implementation, we could achieve meaningful accuracy for classification of the body and context discrepancies, but the accuracy was low for classification of the headline and body discrepancies.

• 한국인공지능학회지
• /
• 제11권4호
• /
• pp.21-27
• /
• 2023

This paper discusses a deep learning-based road surface analysis system that collects data by installing vibration sensors on the 4-axis wheel bearings of a vehicle, analyzes the data, and appropriately classifies the characteristics of the current driving road surface for use in the vehicle's control system. The data used for road surface analysis is real-time large-capacity data, with 48K samples per second, and the A2B protocol, which is used for large-capacity real-time data communication in modern vehicles, was used to collect the data. CAN and CAN-FD commonly used in vehicle communication, are unable to perform real-time road surface analysis due to bandwidth limitations. By using A2B communication, data was collected at a maximum bandwidth for real-time analysis, requiring a minimum of 24K samples/sec for evaluation. Based on the data collected for real-time analysis, performance was assessed using deep learning models such as LSTM, GRU, and RNN. The results showed similar road surface classification performance across all models. It was also observed that the quality of data used during the training process had an impact on the performance of each model.

• Journal of Multimedia Information System
• /
• 제7권2호
• /
• pp.115-124
• /
• 2020

This paper is a study of deep learning among artificial intelligence technology which has been developing many technologies recently. Especially, I am talking about emotional computing that has been mentioned a lot recently during deep learning. Emotional computing, in other words, is a passive concept that is dominated by people who scientifically analyze human sensibilities and reflect them in product development or system design, and a more active concept that studies how devices and systems understand humans and communicate with people in different modes. This emotional signal extraction, sensitivity, and psychology recognition technology is defined as a technology to process, analyze, and recognize psycho-sensitivity based on micro-small, hyper-sensor technology, and sensitive signals and information that can be sensed by the active movement of the autonomic nervous system caused by human emotional changes in everyday life. Chapter 1 talks about overview and Chapter 2 shows related research. Chapter 3 shows the problems and models of real emotional computing and Chapter 4 shows this paper as a conclusion.