• Clinical and Experimental Pediatrics
• /
• 제65권5호
• /
• pp.239-249
• /
• 2022

Cells survive and proliferate through complex interactions among diverse molecules across multiomics layers. Conventional experimental approaches for identifying these interactions have built a firm foundation for molecular biology, but their scalability is gradually becoming inadequate compared to the rapid accumulation of multiomics data measured by high-throughput technologies. Therefore, the need for data-driven computational modeling of interactions within cells has been highlighted in recent years. The complexity of multiomics interactions is primarily due to their nonlinearity. That is, their accurate modeling requires intricate conditional dependencies, synergies, or antagonisms between considered genes or proteins, which retard experimental validations. Artificial intelligence (AI) technologies, including deep learning models, are optimal choices for handling complex nonlinear relationships between features that are scalable and produce large amounts of data. Thus, they have great potential for modeling multiomics interactions. Although there exist many AI-driven models for computational biology applications, relatively few explicitly incorporate the prior knowledge within model architectures or training procedures. Such guidance of models by domain knowledge will greatly reduce the amount of data needed to train models and constrain their vast expressive powers to focus on the biologically relevant space. Therefore, it can enhance a model's interpretability, reduce spurious interactions, and prove its validity and utility. Thus, to facilitate further development of knowledge-guided AI technologies for the modeling of multiomics interactions, here we review representative bioinformatics applications of deep learning models for multiomics interactions developed to date by categorizing them by guidance mode.

• 한국군사과학기술학회지
• /
• 제25권3호
• /
• pp.300-310
• /
• 2022

A vast amount of labeled data is required for deep neural network training. A typical strategy to improve the performance of a neural network given a training data set is to use data augmentation technique. The goal of this work is to offer a novel image augmentation method for improving object detection accuracy. An object in an image is removed, and a similar object from the training data set is placed in its area. An in-painting algorithm fills the space that is eliminated but not filled by a similar object. Our technique shows at most 2.32 percent improvements on mAP in our testing on a military vehicle dataset using the YOLOv4 object detector.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송∙미디어공학회 2021년도 하계학술대회
• /
• pp.23-25
• /
• 2021

This paper describes an approach of composing music with multiple AI composers. This approach enriches more the creativity space of artificial intelligence music composition than using only one composer. This paper presents a simple example with 2 different deep learning composers working together for composing one music. For the experiment, the two composers adopt the same deep learning architecture of an LSTM model trained with different data. The output of a composer is a sequence of notes. Each composer alternatively appends its output to the resulting music which is input to both the composers. Experiments compare different music generated by the proposed multiple composer approach with the traditional one composer approach.

• 천문학회보
• /
• 제43권1호
• /
• pp.54.2-54.2
• /
• 2018

It is important to understand very fast CMEs which are the main cause of geomagnetic storms and solar particle events (SPEs). During this solar cycle 24, there are 10 very fast CMEs whose speeds are over 2000 km/s. Among these, there were only two fronside events (2012 January 23 and 2012 March 7) and they are associated with two major flares (M8.7 and X5.4) and the most strong SPEs (6310 pfu and 6530 pfu). They have a similar characteristics: there were successive CMEs within 2 hours in the same active region. We analyze their magnetic properties using SDO HMI magnetograms and kinematic ones from STEREO EUVI/COR1/COR2 observations. We can measure their speeds and initial accelerations without projection effects because their source locations are almost the limb. Additionally, we are investigating magnetic and kinematic characteristics of 8 backside events using AI-generated magnetograms constructed by deep learning methods.

• The Journal of the Acoustical Society of Korea
• /
• 제14권2E호
• /
• pp.43-49
• /
• 1995

인공지능(AI)의 한기법인 Space Search 기법을 이용하여 회로의 단선 결함의 유무 및 결함위치를 찾아내는 방법을 제시하였다. 보통 회로의 결함은 단선 및 단락의 구조적 결함으로 나뉘어진다. 두가지 결함 모두 회로의 기능에 중대한 이상을 초래한다. 그중 단선에 의한 회로의 결함에 대하여 다루었다. 우선 회로를 net와 net의 연결 path에 따라 tree 구조로 변환하였다. 서로 독립된 net들은 서로 다른 tree의 node를 이루며 각각의 tree는 적기적으로 연결됨이 없다. 각 tree의 최상단부의 root node에 test vector를 입력하고 최하단부의 leaf node에서 vector를 관찰하여 입력된 test vector와 비교한다. 그 비교 결과 동일 유무에 따라 결함의 유무를 판정한다. 결함이 있다고 판정된 leaf node는 depth search 방법에 의하여 root node쪽으로 test vector를 관찰하여, 전기적 신호에 의하여 회로의 서놔 단선된 위치를 찾아내도록 하는 방법을 제시하였다.

• 충청수학회지
• /
• 제8권1호
• /
• pp.19-27
• /
• 1995

In 1974 H.Herrlich invented nearness spaces, a very fruitful concept which enables one to unify topological aspects. In this paper, we introduce the Lindel$\ddot{o}$f nearness structure, countably bounded nearness structure and countably totally bounded nearness structure. And we show that (X, ${\xi}_L$) is concrete and complete if and only if ${\xi}_L={\xi}_t$ in a symmetric topological space (X, t). Also we show that the following are equivalent in a symmetric topological space (X, t): (1) (X, ${\xi}_L$) is countably totally bounded. (2) (X, ${\xi}_t$) is countably totally bounded. (3) (X, t) is countably compact.

• 한국건축시공학회지
• /
• 제7권3호
• /
• pp.91-98
• /
• 2007

Recently a construction industry introduces information that brings about many advantages in the early planning phase, design phase and construction phase. Especially it replaces 2D, 3D systems(usually using explanation of drawing information) ai 4D CAD(offering a sort of 4D-having relation of construction schedule and 3D drawing information). Nevertheless a 4D has these benefits, it has limits which are not only usually using 3D modeling but also limit of making full use of practical affairs because of a lack of connecting varietals of progress of work. To solve these uppermost limits, this research is presenting unified systems to use in risk management which are efficient management of space and non-space information, space analysis, making full use of data base, introducing GIS system of easy interaction.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2003년도 Proceedings of ACRS 2003 ISRS
• /
• pp.1356-1358
• /
• 2003

In the paper, a method making use of the characteristics of Dyad Green Function (DGF) and Fluctuation-Dissipation Theorem to get the brightness temperature of layered medium is introduced. Based on the approach and the measured data of multi-channel radiometer and Least Square Method (LSM), the thickness of lunar soil can be retrieved. These methods are significant to study on materials on lunar surface.

• Journal of Multimedia Information System
• /
• 제9권2호
• /
• pp.93-102
• /
• 2022

Dependency parsing is a decision problem of the syntactic relation between words in a sentence. Recently, deep learning models are used for dependency parsing based on the word representations in a continuous vector space. However, it causes a mislabeled tagging problem for the proper nouns that rarely appear in the training corpus because it is difficult to express out-of-vocabulary (OOV) words in a continuous vector space. To solve the OOV problem in dependency parsing, we explored the proper noun embedding method according to the embedding unit. Before representing words in a continuous vector space, we replace the proper nouns with a special token and train them for the contextual features by using the multi-layer bidirectional LSTM. Two models of the syllable-based and morpheme-based unit are proposed for proper noun embedding and the performance of the dependency parsing is more improved in the ensemble model than each syllable and morpheme embedding model. The experimental results showed that our ensemble model improved 1.69%p in UAS and 2.17%p in LAS than the same arc-eager approach-based Malt parser.

• 한국컴퓨터정보학회논문지
• /
• 제27권3호
• /
• pp.45-52
• /
• 2022

체감형 스포츠 게임은 센서나 카메라를 이용하여 사용자의 신체 움직임을 추적하고 현실감을 느끼게 하는 운동 게임이다. 최근 체감형 스포츠 게임을 학교 체육활동에 사용하기 위해 가상현실 실내 스포츠룸 시스템을 설치하고 있다. 그러나 이러한 시스템은 주로 화면 터치 사용자 상호작용을 사용한다. 본 연구에서는 2차원 벽 터치 인터랙션이 아닌 3차원 공간에서 사용자의 움직임을 추적할 수 있도록 AI 이미지 인식 기술을 사용하는 브이캐치 비전 시스템을 개발했다. 그리고 본 시스템의 운동 효과를 알아보기 위하여 사용성 평가 실험을 진행했다. 실험에서 피험자에게 혈중 산소 포화도와 실시간 심박변이와 키넥트 골격 이동량, 각도 변화량을 측정하여 정량적 운동 효과를 살펴보았다. 실험 결과 브이캐치 비젼 시스템 사용 시 통계적으로 유의미한 심박수 증가와 신체 움직임양 증가로 운동 효과가 있었던 것으로 나타났다. 실험후 설문조사 주관적 평가 결과에서 대부분의 피실험자들은 이 시스템을 사용한 운동이 재미있고 만족스러워했다.