• Journal of Convergence for Information Technology
• /
• v.9 no.8
• /
• pp.1-8
• /
• 2019

Dialog system is becoming a new dominant interaction way between human and computer. It allows people to be provided with various services through natural language. The dialog system has a common structure of a pipeline consisting of several modules (e.g., speech recognition, natural language understanding, and dialog management). In this paper, we tackle a task of domain classification for the natural language understanding module by employing machine learning models such as convolutional neural network and random forest. For our dataset of seven service domains, we showed that the random forest model achieved the best performance (F1 score 0.97). As a future work, we will keep finding a better approach for domain classification by investigating other machine learning models.

• Journal of the Korean Society of Industry Convergence
• /
• v.23 no.6_2
• /
• pp.1081-1091
• /
• 2020

The most recent issue of offshore plants that produce oil and gas are the decommissioning engineering of aged or discontinued platforms. There are many platforms that are being dismantled in the United States, Europe, and areas in Southeast Asia. In particular, more than 400 old platforms in Southeast Asia (Indonesia, Malaysia) are preparing to dismantle. They are spread out across Southeast Asia with a water level of 50 meters and small-scale of less than 10,000 tons. However, this offshore plant decommissioning market is a very suitable market for small and medium-sized shipyards in Korea to enter with their established equipment and engineers. Platform decommissioning is conducted according to decommissioning procedures. However, there are some difficulties in market advances as no developed case studies or process models are established on how platform structures and components are to be dismantled and how the dismantled material is to be reused and recycled. Therefore, this study presented domestic and foreign regulations on the reuse and recycling of oil and gas producing offshore plant platforms, case analyses on developed decommissioning engineering, platform reuse and recycling guidelines, and platform and pipeline decommissioning processes and methods.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.4
• /
• pp.1246-1262
• /
• 2021

Recent studies have demonstrated the strong ability of deep convolutional neural networks (CNNs) to significantly boost the performance in single image super-resolution (SISR). The key concern is how to efficiently recover and utilize diverse information frequencies across multiple network layers, which is crucial to satisfying super-resolution image reconstructions. Hence, previous work made great efforts to potently incorporate hierarchical frequencies through various sophisticated architectures. Nevertheless, economical SISR also requires a capable structure design to balance between restoration accuracy and computational complexity, which is still a challenge for existing techniques. In this paper, we tackle this problem by proposing a competent architecture called Enhanced U-Net Network (EUN), which can yield ready-to-use features in miscellaneous frequencies and combine them comprehensively. In particular, the proposed building block for EUN is enhanced from U-Net, which can extract abundant information via multiple skip concatenations. The network configuration allows the pipeline to propagate information from lower layers to higher ones. Meanwhile, the block itself is committed to growing quite deep in layers, which empowers different types of information to spring from a single block. Furthermore, due to its strong advantage in distilling effective information, promising results are guaranteed with comparatively fewer filters. Comprehensive experiments manifest our model can achieve favorable performance over that of state-of-the-art methods, especially in terms of computational efficiency.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.307-312
• /
• 2022

상호참조해결은 주어진 문서에서 멘션을 추출하고 동일한 개체의 멘션들을 군집화하는 작업이다. 기존 상호참조해결 연구의 멘션탐지 단계에서 진행한 가지치기는 모델이 계산한 점수를 바탕으로 순위화하여 정해진 비율의 멘션만을 상호참조해결에 사용하기 때문에 잘못 예측된 멘션을 입력하거나 정답 멘션을 제거할 가능성이 높다. 또한 멘션 탐지와 상호참조해결을 종단간 모델로 진행하여 학습 시간이 오래 걸리고 모델 복잡도가 높은 문제가 존재한다. 따라서 본 논문에서는 상호참조해결을 2단계 파이프라인 모델로 진행한다. 첫번째 멘션 탐지 단계에서 후보 단어 범위의 점수를 계산하여 멘션을 예측한다. 두번째 상호참조해결 단계에서는 멘션 탐지 단계에서 예측된 멘션을 그대로 이용해서 서로 상호참조 관계인 멘션 쌍을 예측한다. 실험 결과, 2단계 학습 방법을 통해 학습 시간을 단축하고 모델 복잡도를 축소하면서 종단간 모델과 유사한 성능을 유지하였다. 상호참조해결은 Light에서 68.27%, AMI에서 48.87%, Persuasion에서 69.06%, Switchboard에서 60.99%의 성능을 보였다.

• Annual Conference of KIPS
• /
• 2022.05a
• /
• pp.549-552
• /
• 2022

최근 소셜미디어 리뷰 데이터를 활용한 약물 이상 반응 탐지 연구가 활발히 진행되고 있지만, 약물을 복용하기 전 증상과 약물 이상 반응을 구분하지 못한다는 한계가 있다. 본 논문에서는 약물 이상 반응 탐지에서 약물 복용 전의 증상을 구분할 수 있는 Filtering Clinical BERT(FC-BERT) 모델을 제안하였다. FC-BERT 는 약물 복용 전 증상과 다른 약물에 대한 부작용 표현을 제거하기 위해 약물명이 나오기 전 모든 문장을 제거하는 필터링과 약물-부작용 쌍을 추출하는 모델을 사용했다. 성능 평가 실험을 위해 문장에 대한 ADE(Adverse Drug Event) 여부가 들어있는 ADE Corpus V2 데이터를 활용하였고 SPARK NLP 라이브러리에서 제공하는 ADE Pipeline 모델과 비교하여 성능 평가를 실시하였다. 실험 결과 필터링을 활용한 FC-BERT 모델이 기존 모델보다 정확도, 평균 정밀도, 평균 재현율, 평균 F1-score 가 모두 높은 결과를 보여주었다. 본 논문에서 제시한 모델은 기존 연구의 한계점을 보완하여 보다 정확한 약물 부작용 시그널을 탐지하는데 기여할 수 있을 것이다.

• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.1
• /
• pp.54.2-54.2
• /
• 2021

Galaxy clusters are the largest structures in the universe located at the top of the cosmological hierarchical model, so the evolution of the universe can be understood by studying clusters of galaxies. Therefore, finding a larger number of galaxy clusters plays an important role in exploring how the universe evolves. A large number of catalogs for galaxy clusters in the northern sky have been published; however, there are few catalogs in the southern sky due to the lack of wide sky survey data. KMTNet Synoptic Survey of Southern Sky(KS4) project, which observes a wide area of the southern sky about 7000 deg² with KMTNet telescopes for two years, is in progress under the SNU Astronomy Research Center. We use the KS4 multi-wavelength optical data and measure photometric redshifts of galaxies for finding galaxy clusters at redshift z<1. Currently, the KS4 project has observed approximately 33% of the target region, and a pipeline that measures photometric redshifts of galaxies has been created. When the project is completed, we expect to find more than a hundred thousand galaxy clusters, and this will improve the study of galaxy clusters in the southern sky.

• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.2
• /
• pp.68.2-68.2
• /
• 2021

Galaxy clusters are the largest structures in the universe located at the top of the cosmological hierarchical model, so the evolution of the universe can be understood by studying clusters of galaxies. Therefore, finding a larger number of galaxy clusters plays an important role in exploring how the universe evolves. A large number of catalogs for galaxy clusters in the northern sky have been published; however, there are few catalogs in the southern sky due to the lack of wide sky survey data. KMTNet Synoptic Survey of Southern Sky(KS4) project, which observes a wide area of the southern sky about 7000 deg² with KMTNet telescopes for two years, is in progress under the SNU Astronomy Research Center. We use the KS4 multi-wavelength optical data and measure photometric redshifts of galaxies for finding galaxy clusters at redshift z<1. Currently, the KS4 project has observed approximately 50% of the target region, and a pipeline that measures photometric redshifts of galaxies has been created. When the project is completed, we expect to find more than a hundred thousand galaxy clusters, and this will improve the study of galaxy clusters in the southern sky.

• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.264-269
• /
• 2021

상호참조해결은 주어진 문서에서 상호참조해결의 대상이 될 수 있는 멘션을 추출하고, 같은 개체를 의미하는 멘션 쌍 또는 집합을 찾는 자연어처리 작업이다. 하나의 멘션 내에 멘션이 될 수 있는 다른 단어를 포함하는 중첩 멘션은 순차적 레이블링으로 해결할 수 없는 문제가 있다. 본 논문에서는 이러한 문제를 해결하기 위해 멘션의 시작 단어의 위치를 여는 괄호('('), 마지막 위치를 닫는 괄호(')')로 태깅하고 이 괄호들을 예측하는 멘션 탐지 모델과 멘션 탐지 모델에서 예측된 멘션을 바탕으로 포인터 네트워크를 이용하여 같은 개체를 나타내는 멘션을 군집화하는 상호참조해결 모델을 제안한다. 실험 결과, 4개의 영어 대화 데이터셋에서 멘션 탐지 모델은 F1-score (Light) 94.17%, (AMI) 90.86%, (Persuasion) 92.93%, (Switchboard) 91.04%의 성능을 보이고, 상호참조해결 모델에서는 CoNLL F1 (Light) 69.1%, (AMI) 57.6%, (Persuasion) 71.0%, (Switchboard) 65.7%의 성능을 보인다.

• Nuclear Engineering and Technology
• /
• v.55 no.11
• /
• pp.4057-4065
• /
• 2023

Organic scintillators are widely used for neutron/gamma detection. Pulse shape discrimination algorithms have been commonly used to discriminate the detected radiations. These algorithms have several limits, in particular with plastic scintillator which has lower discrimination ability, compared to liquid scintillator. Recently, machine learning (ML) models have been explored to enhance discrimination performance. Nevertheless, obtaining an accurate ML model or evaluating any discrimination approach requires a reference neutron dataset. The preparation of this is challenging because neutron sources are also gamma-ray emitters. Therefore, this paper proposes a pipeline to prepare clean labeled neutron/gamma datasets acquired by an organic scintillator. The method is mainly based on a Time of Flight setup and Tail-to-Total integral ratio (TTT_ratio) discrimination algorithm. In the presented case, EJ276 plastic scintillator and ²⁵²Cf source were used to implement the acquisition chain. The results showed that this process can identify and remove mislabeled samples in the entire ToF spectrum, including those that contribute to peak values. Furthermore, the process cleans ToF dataset from pile-up events, which can significantly impact experimental results and the conclusions extracted from them.

• Journal of Urban Science
• /
• v.12 no.2
• /
• pp.53-64
• /
• 2023

With the development of the plant industry, there has been an increasing frequency of major accidents both domestically and internationally, emphasizing the importance of plant safety. Therefore, this study aims to investigate measures to enhance the stability of piping, a key component within the plant. Upon examining the piping, erosion, buckling, and fatigue emerged as significant risk factors among various potential hazards, leading to their selection as the primary risk factors in this study. Identifying variables that can collectively mitigate these factors, the study focuses on the material, thickness, and elbow angle of the piping. The reference piping model is established as the pipeline connecting the Skim Tank and IGF within a 300BPD oil sands modular plant in Yeoncheon, Gyeonggi-do. Utilizing the FEA analysis program ANSYS, the study conducts a variable analysis for the identified risk factors. The results of the analysis, through comparison and evaluation, provide evidence of the effectiveness of enhancing stability. It is observed that reducing the elbow angle significantly improves erosion and buckling, while changing to a material with high yield stress most significantly enhances stability when considering fatigue.