• Journal of KIISE
• /
• v.44 no.1
• /
• pp.36-43
• /
• 2017

Syntactic information represents the dependency relation between predicates and arguments, and it is helpful for improving the performance of Semantic Role Labeling systems. However, syntax analysis can cause computational overhead and inherit incorrect syntactic information. To solve this problem, we exclude syntactic information and use only morpheme information to construct Semantic Role Labeling systems. In this study, we propose an end-to-end SRL system that only uses morpheme information with Stacked Bidirectional LSTM-CRFs model by extending the LSTM RNN that is suitable for sequence labeling problem. Our experimental results show that our proposed model has better performance, as compare to other models.

• Journal of Digital Contents Society
• /
• v.18 no.7
• /
• pp.1249-1257
• /
• 2017

Due to the 4th industrial revolution and an aged society, many studies are being conducted to apply virtual reality to medical field. Research on dementia is especially active. This paper proposes virtual reality based on cognitive rehabilitation contents using image recognition and clustering method to improve cognitive and physical disabilities caused by dementia. Unlike the existing cognitive rehabilitation system, this paper uses travel photos that reflect the memories of the subjects to be treated. In order to generate automated cognitive rehabilitation contents, we extract face information, food pictures, place information, and time information from photographs, and normalization is performed for clustering. And we present scenarios that can be used as cognitive rehabilitation contents using travel photos in virtual reality space.

• The Journal of Korea Robotics Society
• /
• v.13 no.1
• /
• pp.38-44
• /
• 2018

In this paper, we propose Intelligent Driver Assistance System (I-DAS) for driver safety. The proposed system recognizes safety and danger status by analyzing blind spots that the driver cannot see because of a large angle of head movement from the front. Most studies use image pre-processing such as face detection for collecting information about the driver's head movement. This not only increases the computational complexity of the system, but also decreases the accuracy of the recognition because the image processing system dose not use the entire image of the driver's upper body while seated on the driver's seat and when the head moves at a large angle from the front. The proposed system uses a convolutional neural network to replace the face detection system and uses the entire image of the driver's upper body. Therefore, high accuracy can be maintained even when the driver performs head movement at a large angle from the frontal gaze position without image pre-processing. Experimental result shows that the proposed system can accurately recognize the dangerous conditions in the blind zone during operation and performs with 95% accuracy of recognition for five drivers.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.15 no.2
• /
• pp.71-78
• /
• 2020

This paper proposes vision sensors and deep learning-based around view monitoring system for ship berthing. Ship berthing to the port requires precise relative position and relative speed information between the mooring facility and the ship. For ships of Handysize or higher, the vesselships must be docked with the help of pilots and tugboats. In the case of ships handling dangerous cargo, tug boats push the ship and dock it in the port, using the distance and velocity information receiving from the berthing aid system (BAS). However, the existing BAS is very expensive and there is a limit on the size of the vessel that can be measured. Also, there is a limitation that it is difficult to measure distance and speed when there are obstacles near the port. This paper proposes a relative distance and speed estimation system that can be used as a ship berthing assist system. The proposed system is verified by comparing the performance with the existing laser-based distance and speed measurement system through the field tests at the actual port.

• Proceedings of the Korea Technology Innovation Society Conference
• /
• 2017.11a
• /
• pp.715-721
• /
• 2017

기술거래 시장의 활성화에 대한 연구개발서비스 분야 종사자들의 관심이 높아지고 있으며, 특히 공공 및 민간 분야의 휴면 기술(특허)에 대한 이전 거래를 통해 불필요한 특허유지 비용을 줄이고 부가적인 기술료 창출 효과를 거둘 수 있다. 본 연구에서는 현재까지 기술이전(거래), 현물출자, 기술금융(융자, 담보대출) 등 다양한 목적으로 실무에서 활용되어 온 기술가치평가 모형의 한계점을 고민해 보고, 이에 대한 개선방안으로서 몬테카를로 최소자승법 기반의 확률론적 가치평가 모형을 제시한다. 기존의 가치평가 모형은 평가산출을 위한 입력변수의 확정적 값들에 기반하여 가치액이 산출되었으나, 대표적 기법인 현금흐름 할인법이나 로열티공제법의 경우 미래의 수익예상기간, 예상매출액 등에서는 불확실성(uncertainty)가 내재되어 있다. 따라서 특정 분포(distribution)에 대한 확률론적 가능성을 가정하고 이에 대한 수학적 최적화 논리로부터 몬테카를로 최소자승 관게에 의한 변수결정 및 가치평가액 산정을 할 수 있는 평가모듈을 개발한다. 향후 연구에서는 기 평가된 사례결과를 딥러닝(deep learning) 방식으로 학습하여, 발생가능성 높은 각 변수값의 범위들을 산출하고 이로부터 기술가치 범위를 추론하는 시스템을 개발하는 것도 가능할 것으로 기대된다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2018.10a
• /
• pp.342-344
• /
• 2018

This paper shows the possibility to substitute statistical methods such as mean imputation, correlation coefficient analysis, graph correlation analysis for the proposed algorithm, and replace statistician for processing various abnormal data measured in the water treatment process with it. In addition, this study aims to model a data-filtering system based on a recent fractile pattern and a deep learning-based LSTM algorithm in order to improve the reliability and validation of the algorithm, using the open-sourced libraries such as KERAS, THEANO, TENSORFLOW, etc.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.29 no.6
• /
• pp.1393-1401
• /
• 2019

Recently as voice authentication function is installed in the system, it is becoming more important to accurately authenticate speakers. Accordingly, a model for verifying speakers in various ways has been suggested. In this paper, we propose a new method for verifying speaker verification using a Short-time Fourier Transform(STFT). Unlike the existing Mel-Frequency Cepstrum Coefficients(MFCC) extraction method, we used window function with overlap parameter of around 66.1%. In this case, the speech characteristics of the speaker with the temporal characteristics are studied using a deep running model called RNN (Recurrent Neural Network) with LSTM cell. The accuracy of proposed model is around 92.8% and approximately 5.5% higher than that of the existing speaker certification model.

• Annual Conference of KIPS
• /
• 2019.10a
• /
• pp.946-949
• /
• 2019

한 벌의 한글 글자체를 만드는데 일반적으로 많은 제작 비용과 시간이 소요된다. 따라서 폰트 제작의 어려움을 덜기 위해, 사용자가 대표 글자들을 입력하면 그 글자들의 디자인 특성을 딥러닝 기술을 이용하여 학습한 모델이 나머지 글자들을 자동 생성해주는 시스템 구축한다면 폰트 제작이 훨씬 용이해질 뿐만 아니라 저작권 문제로부터 자유로워질 것이다. 이와 관련된 선행연구를 실행하고 분석해 본 결과 데이터 전처리 과정에서 글자가 잘리거나 크기가 맞지 않아 제대로 된 데이터셋이 구축되지 않는 문제가 있음을 발견하였다. 본 논문에서는 이러한 문제를 해결하기 위해 템플릿에서 자동적으로 글자영역을 추출하고 이미지를 보정하는 전처리 과정과 함께 기존 모델에서 새로운 필터를 추가하여 학습 성능을 높이는 방법을 제안한다. 이를 통해 기존 연구에서 측정된 손실값을 낮춘 결과를 확인했으며 결과적으로 실제 글자체와 더욱 유사한 사용자 맞춤형 글자체를 제공할 수 있을 것이다.

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.17 no.4
• /
• pp.53-62
• /
• 2021

Deep learning models such as convolutional neural networks and recurrent neual networks process a huge amounts of data, so they require a lot of storage and consume a lot of time and power due to memory access. Recently, research is being conducted to reduce memory usage and access by compressing data using the feature that many of deep learning data are highly sparse and localized. In this paper, we propose a compression-decompression method of storing only the non-zero data and the location information of the non-zero data excluding zero data. In order to make the location information of non-zero data, the matrix data is divided into sections uniformly. And whether there is non-zero data in the corresponding section is indicated. In this case, section division is not executed only once, but repeatedly executed, and location information is stored in each step. Therefore, it can be properly compressed according to the ratio and distribution of zero data. In addition, we propose a hardware structure that enables compression and decompression without complex operations. It was designed and verified with Verilog, and it was confirmed that it can be used in hardware deep learning accelerators.

• Journal of Korea Multimedia Society
• /
• v.24 no.2
• /
• pp.233-244
• /
• 2021

Recently, several ICT-based cat studies have produced some successful results, according to academic and industry sources. However, research on the level of simply identifying the cat's condition, such as the behavior and sound classification of cats based on images and sound signals, has yet to be found. In this paper, based on the veterinary scientific knowledge of cats, a practical and academic cat monitoring and disease diagnosis system is proposed to monitor the health status of the cat 24 hours a day by automatically categorizing and analyzing the behavior of the cat with location information using LSTM with a beacon sensor and a raspberry pie that can be built at low cost. Validity of the proposed system is verified through experimentation with cats in actual custody (the accuracy of the cat behavior classification and location identification was 96.3% and 92.7% on average, respectively). Furthermore, a rule-based disease analysis system based on the veterinary knowledge was designed and implemented so that owners can check whether or not the cats have diseases at home (or can be used as an auxiliary tool for diagnosis by a pet veterinarian).