• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2017.11a
• /
• pp.88-91
• /
• 2017

넓은 동적 영역 (high dynamic range: HDR) 이미지는 시각적으로 우수하지만 대부분의 디스플레이는 좁은 동적 영역 (low dynamic range: LDR)만 지원이 가능하다. 이를 해결하기 위해서 톤 매핑 기법 (tone mapping operator: TMO)을 사용한 동적 영역 압축을 수행한다. 기존의 적응적 로그 매핑 (adaptive logarithmic mapping)의 경우 에지 부분에서 디테일이 손실되는 문제점이 있었다. 본 논문에서는 가이디드 이미지 필터링 (guided image filtering: GIF)을 통해 베이스 레이어와 디테일 레이어로 나눠서 처리하는 알고리듬을 제안한다. 베이스 레이어는 적응적 로그 매핑을 통해 동적 영역을 압축하고 디테일 레이어와 더해 기존의 톤 매핑 과정에서 발생하는 디테일의 손실을 감소시켰다.

• Journal of KIISE:Computing Practices and Letters
• /
• v.8 no.2
• /
• pp.127-140
• /
• 2002

In a spatial database system, the semantic integrity should be supported for maintaining the data consistency. In the real world, spatial objects In boundary layer should always meet neighbor objects, and they cannot hold the same name. This characteristic is an implied concept in real world. So, when this characteristic is disobeyed due to the update operations of spatial objects, it is necessary to maintain the integrity of a layer. In this thesis, we propose a spatial-operation-trigger for supporting the integrity of spatial objects. The proposed method is defined a spatial-operation-trigger based on SQL-3 and executed when the constraint condition is violated. A spatial-operation-trigger have the strategy of execution. Firstly, for one layer, the spatial and aspatial data triggers are executed respectively. Secondly, the aspatial data trigger for the other layers is executed. Spatial-operation-trigger for one layer checks whether the executed operation updates only spatial data, aspatial data, or both of them, and determines the execution strategy of a spatial-operation-trigger. Finally, the aspatial data trigger for the other layers is executed. A spatial-operation-trigger is executed in three steps for the semantic integrity of the meet-property of spatial objects. And, it provides the semantic integrity of spatial objects and the convenience for users using automatic correcting operation.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2017.06a
• /
• pp.133-134
• /
• 2017

본 논문에서는 차세대 방송 표준 ATSC (Advanced Television Systems Committee) 3.0 기반 LDM (Layered Division Multiplexing) 및 스케일러블 비디오 코덱 (Scalable Video Codec) 을 활용한 지상파 방송시스템 기술을 살펴보고 그 성능을 비교 분석한다. 코어 레이어 (Core Layer)와 인핸스드 레이어 (Enhanced Layer)로 구성된 LDM 기반 PLP (Physical Layer Pipe)에, 스케일러블 비디오 코딩이 적용된 베이스 레이어 (Base Layer)와 인핸스먼트 레이어 (Enhancement Layer) 스트림을 각각 전송함으로써 하나의 RF 채널에 두 개 이상의 서비스를 전달할 경우 채널 효율을 극대화 할 수 있다. 본 논문에서는 이동 및 고정용 서비스, 즉 두 개의 서비스를 전송할 때 제안된 LDM 및 스케일러블 비디오 코덱을 사용한 기술과 TDM (Time Division Multiplexing) 및 Simulcast 를 적용한 기술과의 성능 비교를 통해 제안된 기술의 우수성을 검증하고자 한다.

• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.111-114
• /
• 2019

BERT는 자연어처리 여러 응용 분야(task)에서 우수한 성능을 보여줬으나, BERT 사전학습 모델을 학습하기 위해서는 많은 학습 시간과 학습 자원이 요구된다. 본 논문에서는 빠른 학습을 위한 한국어 BERT 학습 방법을 제안한다. 본 논문에서는 다음과 같은 세 가지 학습 방법을 적용했다. 교착어인 한국어 특성을 반영하기 위해 형태소 분석 기반의 사전을 사용하였으며, 단어 간 상대적 위치 표현을 추가하여, 상대적 위치 정보를 학습했다. 또한 BERT 베이스 모델의 12-레이어 중 3-레이어만을 사용하여, 모델을 경량화시켰다.

• Journal of KIISE:Databases
• /
• v.35 no.2
• /
• pp.132-142
• /
• 2008

In biological domain, there exist a variety of protein analysis databases which have their own meaning toward the same target of protein. If we integrate these scattered heterogeneous data efficiently, we can obtain useful information which otherwise cannot be found from each original source. Reflecting the characteristics of biological data, each data source has its own syntax and semantics. If we describe these data through RDF(S) models, one of the Semantic Web standards, we can achieve not only syntactic but also semantic integration. In this paper, we propose a new concept of integration layer based on the RDF unified schema. As a conceptual model, we construct a unified schema focusing on the protein information; as a representational model, we propose a technique for the wrappers to aggregate necessary information from the relevant sources and dynamically generate RDF instances. Two example queries show that our integration layer succeeds in processing the integrated requests from users and displaying the appropriate results.

• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.465-469
• /
• 2018

본 논문에서는 지식베이스 완성을 위한 새로운 모델, KBCNN을 소개한다. KBCNN 모델은 CNN을 기반으로 지식베이스의 개체들과 관계들 사이의 연관성을 포착한다. KBCNN에서 각 트리플 <주어 개체, 관계, 목적어 개체>는 3개의 열을 가진 행렬로 표현되며, 각각의 열은 트리플의 각 원소를 표현하는 임베딩 벡터다. 트리플을 나타내는 행렬은 여러 개의 필터를 가지고 있는 컨볼루션 레이어를 통과한 뒤, 하나의 특성 벡터로 합쳐진다. 이 특성 벡터를 가중치 행렬과 내적 하여 최종적으로 해당 트리플의 신뢰도를 출력하게 된다. 이 신뢰도를 바탕으로 트리플의 진실 여부를 가려낼 수 있다. 지식베이스 완성 연구에서 가장 많이 사용되는 데이터셋인 FB15k-237을 기반으로 한 실험을 통해 KBCNN 모델이 기존 임베딩 모델들보다 뛰어난 성능을 보이는 것을 확인하였다.

• Journal of the Korean Society of Clothing and Textiles
• /
• v.41 no.5
• /
• pp.872-882
• /
• 2017

There are no specific guidelines on how to control the heat input for the heat generating smart base layer. This study investigated the mode of actuating heat pad attached to the base layer by performing a human wear test in a cold environment. Subjects participating in the test wore T-shirts, jumper and pants on the base layer with heating pads. Skin temperature, total time of heating and the number of switching for the heating mode were observed as the subject controlled the heating mode voluntarily. The comfortable range of skin temperature on the abdomen was larger than on the lower back. The subject felt hot and turned off the switch when the mean skin temperature of the abdomen was $48.8^{\circ}C$ and the lower back was $40.1^{\circ}C$. The total heating time and the number of actuating switching were larger for women than men. The voluntary action of heating for men with high cold sensitivity was significantly different from men with low cold sensitivity. Therefore, it is necessary (depending on gender and cold sensitivity) to set the heating mode differently for the automatic heat control of a future smart base layer.

• Journal of the Korean Society of Clothing and Textiles
• /
• v.41 no.4
• /
• pp.687-697
• /
• 2017

To figure out an appropriate pressure level for a body warming base layer, human responses were observed when the pattern reduction of base layers varied. Under the condition of $2^{\circ}C$, 60% RH, 0.1m/s, ten male subjects participated in the experiment with four sizes of experimental vests where heating pads were attached. The subjective evaluations of the heating vests with different sizes were reported using 7 or 9 point scales. We simultaneously observed chest, abdomen and scapula skin temperatures and microclimate humidity. It was found that the tight pattern as in the case of A or B provided a warmer subjective sensation and skin temperature than C or D; however, there were no differences in skin temperature at the chest. Eventually, the chest temperature decreased after about 30 mins of heating; however, temperature of abdomen increased and indicated that heating with two commercial pads used was inadequate for whole body warming. The pressure sensation of 'tight' was improved after warming the abdomen in a cold environment. Overall, the gaps beyond the original circumference of the abdomen, as in C or D, were not desirable for the local heating of abdomen under the conditions of this experiment where walking was included in the protocol. The experiment garment B with nude waist circumference was the best, and D with the largest ease, was the worst for a comfortable warming vest.

• Journal of Broadcast Engineering
• /
• v.23 no.4
• /
• pp.474-483
• /
• 2018

In this paper, we propose a multi-scale high dynamic range (HDR) tone mapping algorithm using guided image filter (GIF). The GIF is used to divide an image into a base layer and a detail layer, then the range of the detail layer is reduced with a compression function to enhance the detail information of the image. However, in most cases, an image includes the detail and edge information in different scales. That is to say, it is difficult to represent all detail features under a certain scale, and a single-scale image decomposition method is not free from artifacts around edges. To solve the problems, the multi-scale image decomposition method is proposed. It utilizes the detail layers of several scale to determine how much edge is preserved. Experiment results show that the proposed algorithm has better image performance in preserving edge compared to conventional algorithm.

• Journal of KIISE:Databases
• /
• v.30 no.6
• /
• pp.585-592
• /
• 2003

In the spatial database systems, it is necessary to manage spatial objects that have two or more aspatial information with different security levels on the same layer. If we adapt the polyinstantiation concept of relational database system for these spatial objects, it is difficult to process the representation problem of spatial objects and to solve the security problem that is service denial and information flow by access of subject that has a different security level. To address these problems, we propose a polyinstantiation method for security management of spatial objects in this paper. The proposed method manages secure spatial database system efficiently by creating spatial objects according to user's security level through security-level-conversion-step and polyinstantiation-generation-step with multi-level security policy. Also, in case of user who has a different security level requires secure operations, we create polyinstance for spatial object to solve problems of service denial and information flow.