• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.588-591
• /
• 2021

다중추론 질의응답 태스크는 하나의 문서만 필요한 기존의 단일추론 질의응답(Single-hop QA)을 넘어서 복잡한 추론을 요구하는 질문에 응답하는 것이 목표이다. IRQA에서는 검색 모델의 역할이 중요한 반면, 주목받고 있는 Dense Retrieval 모델 기반의 다중추론 질의응답 검색 모델은 찾기 어렵다. 본 논문에서는 검색분야에서 좋은 성능 보이고 있는 Dense Retrieval 모델의 다중추론을 위한 사전학습 방법을 제안하고 관련 한국어 데이터 셋에서 이전 방법과의 성능을 비교 측정하여 학습 방법의 유효성을 검증하고 있다. 이를 통해 지식 베이스, 엔터티 링킹, 개체명 인식모듈을 비롯한 다른 서브모듈을 사용하지 않고도 다중추론 Dense Retrieval 모델을 학습시킬 수 있음을 보였다.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.44 no.2
• /
• pp.106-111
• /
• 1999

To investigate the effects of planting density and nitrogen level on growth and yield potential of newly bred heavy panicle japonica rice with large grain (Iksan 435 and Iksan 438) or many spikelets per panicle(HR14022-21-8-4 and HR14022-21-8-6), four heavy panicle type rices and two many panicle type rices(Dongjinbyeo and Donganbyeo) as the checks were planted under standard planting density (30$\times$15 cm) and dense planting density (15$\times$15 cm) with two nitrogen levels of standard nitrogen level(110 kg h $a^{-1}$) and heavy nitrogen level(165 kg h $a^{-1}$). Effective tiller rate decreased in dense planting or heavy nitrogen, when compared to standard nitrogen and planting, while leaf area index and to dry weight increased in dense planting or heavy nitrogen. Tiller numbers and panicle numbers were more increased by dense planting than heavy nitrogen, whereas spikelet numbers were more increased by heavy nitrogen than dense planting. Ripened grain ratio was slightly lower only in dense planting. 1,000 grain weight in brown rice was not significantly different in dense planting or heavy nitrogen. Milled rice yield was highest in heavy nitrogen with standard planting for heavy panicle type rice, while yield for many panicle type rice was highest in heavy nitrogen with dense planting, suggesting that many panicle type rice possesses higher adapt-ability for dense planting than heavy panicle type rice. Path coefficient analysis revealed that top dry weight, spikelet number and grain weight were the greatest positive contributors to yield, whereas tiller number was negative to yield.d.

• Korean Journal of Remote Sensing
• /
• v.37 no.5_1
• /
• pp.1061-1070
• /
• 2021

In this manuscript, we tried to improve the performance of the FC-DenseNet by applying an attention gate for the classification of cropping areas. The attention gate module could facilitate the learning of a deep learning model and improve the performance of the model by injecting of spatial/spectral weights to each feature map. Crop classification was performed in the onion and garlic regions using a proposed deep learning model in which an attention gate was added to the skip connection part of FC-DenseNet. Training data was produced using various PlanetScope satellite imagery, and preprocessing was applied to minimize the problem of imbalanced training dataset. As a result of the crop classification, it was verified that the proposed deep learning model can more effectively classify the onion and garlic regions than existing FC-DenseNet algorithm.

• The Korean Journal of Ceramics
• /
• v.5 no.1
• /
• pp.25-29
• /
• 1999

Complex viscosities of dense alumina slurries over 45% volume density measured with the oscillating method were correlated well with Casson model. Among several monosaccharides and disaccharides studied here, fructose and sucrose showed good rheological properties in making dense alumina slurry plastic compared to other monosaccharides and disaccharides like glucose, galactose, arabinose, xylose and maltose. Sucrose content or additional water content in dense alumina slurry with sucrose contributed to the plasticity of the slurries.

• The Bulletin of The Korean Astronomical Society
• /
• v.39 no.2
• /
• pp.75.2-75.2
• /
• 2014

CS molecule as an important tracer for studying inward motions in dense cores is known to be adsorbed onto dusts in cold (T~10K) dense cores, resulting in its significant depletion in the central region of the cores which may hamper a proper study of kinematics stage of star formation. In this study we choose five 'evolved' dense starless cores, L1544, L1552, L1689B, L694-2 and L1197, to investigate how depletion of CS molecule is significant and how the molecule differentiates depending on the evolutional status of the dense cores, by using a rare isotopomer C34S. We performed mapping observations in C34S (J=2-1) and N2H+ (J=1-0) with Nobeyama 45 m telescope, and compared $850{\mu}m$ continuum data as a reference of the density distribution of the dense cores. Our data confirm the claim that CS molecule generally depletes out in the central region in dense starless cores, while N2H+ keeps abundant as they get evolved. All of integrated intensity maps show 'semi-ring-like' depletion holes in CS, and all of abundance radial profiles show decrease toward center. The CS depletion and molecular chemical differentiation seems to depend on the evolutional status in dense cores. The evolved cores shows low abundance at both central and outer regions, implying that in the case of highly evolved cores CS freeze-out occurs over the most area of the cores.

• Journal of Korean Society for Atmospheric Environment
• /
• v.21 no.5
• /
• pp.537-545
• /
• 2005

A new dispersion model for dense gas is constructed in the Lagrangian framework. Prediction of concentration by the proposed model is compared with measure data obtained in the experiment conducted in Thorney Island in 1984. Two major effects of dense gas dispersion, gravity slumping and stratification effect, are successfully incorporated into LDM (Lagrangian dense gas model). Entrainment effect is naturally modelled by introducing stochastic dispersion model with the effect of turbulence suppression by stratification. Not only various releasing conditions but also complex terrain can be extended to, although proposed model is appropriate for flat terrain.

• Proceedings of the KIEE Conference
• /
• 2006.10c
• /
• pp.148-150
• /
• 2006

In this paper, we propose a new method for detection of mass-type breast cancer in dense mammogram. As the proposed method analyzes texture of the breast tissue using method by fusing Homogeneity and Ranklets, improve problem of traditional method. Homogeneity gives the measure of uniform density, and Ranklets determine orientation selective property at vertical, horizontal and diagonal in mass region. The proposed method is suitable to dense mammogram with tangled normal tissue and cancer tissue. SVM(Support Vector Machine) classifier is used for effective detection of mass-type breast cancer in dense mammogram.

• Journal of Mechanical Science and Technology
• /
• v.17 no.7
• /
• pp.1083-1094
• /
• 2003

This study represents an assessment of the phase-Doppler technique to the measurements of dense droplet laden jet. High-pressure injection fuel sprays have been investigated to evaluate the use of the Phase-Doppler anemometry (PDA) technique. The critical issue is the stability of the phase-Doppler anemometry technique for dense droplet laden jet such as Diesel fuel spray in order to insure the results from the drop size and velocity measurements are repeatable, consistent, and physically realistic because the validation rate of experimental data is very low due to the thick optical density. The effect of shift frequency is minor, however, the photomultiplier tube (PMT) voltage setting is very sensitive to the data acquisition and noise in dense droplet laden jet. The optimum PMT voltage and shift frequency should be chosen so that the data such as volume flux and drop diameter do not change rapidly.

• The Korean Journal of Ceramics
• /
• v.5 no.2
• /
• pp.104-109
• /
• 1999

Rheological properties of dense slurries over 45 volume % with different monosaccharides and disaccharides were checked in order to increase the solid content of dense slurries without sacrificing plasticity using creep testing method. Strain in creep test showed good correlations with Burger model which is expressed as an exponential function of time. Among several monosaccharides and disaccharides studied here, fructose and sucrose were most effective in making dense alumina slurry plastic than other monosaccharides and disaccharides like glucose, galactose, xyloss and maltose. In the case of dense alumina slurry with sucrose, sucrose content or additional water content enhanced to the plasticity of the slurries.

• The Bulletin of The Korean Astronomical Society
• /
• v.45 no.1
• /
• pp.42.1-42.1
• /
• 2020

Filamentary structures pervade the whole kind of molecular clouds from low- to high-mass star-forming clouds, and the non-star-forming clouds. It is supposed to be a prerequisite stage of star formation, and hence how filaments and dense cores form is one of the critical questions in the early star formation study. We investigated the dynamics and chemistry of dense cores in IC5146 using TRAO FUNS (TRAO Survey of the nearby Filamentary molecular clouds, the Universal Nursery of Stars) data. In addition, we performed polarization observation using JCMT Pol-2 polarimetry to investigate the magnetic field morphology within a core-scale. In the presentation, we will present the result of TRAO FUNS and JCMT/Pol2 observation toward the filaments and dense cores in the IC5146. We aim to reveal the roles of gravity, turbulence, and magnetic field in the formation of dense cores in the western hub-filament structure of IC5146.